In recent years, more and more terms that sound highly technical have emerged in the lighting industry: CAF, CS, EML, m-EDI, EDI, DER…

Many manufacturers, designers, consultants, and system providers have heard of them—and to some extent, used them. But if we push one step further and ask:

• What exactly does each of these metrics describe?
• Are they actually talking about the same thing?
• Can they be used interchangeably?

The answers are often far less clear than people assume. This reflects a very typical situation in today’s lighting industry: There are more and more terms, but a true common language has not yet been established.

If the industry genuinely wants to move from simply “lighting up spaces” to accurately understanding how light affects people, then the first step is not to invent yet another new term.

It is to put these commonly used metrics back into their proper context.

---

Are these metrics really describing the same thing?

Let’s start with the conclusion: CAF, CS, EML, and EDI/DER are not different names within the same framework.

They originate from:

• different stages
• different objectives
• different modeling approaches

Some function more like spectral efficacy ratios.
Some behave more like physiological response models.
Some are closer to application-level compromise metrics.
And others are more like standardized, computable, and transferable baseline coordinates.

---

So the issue is not whether these terms should exist

The real issue is this: If the industry treats all of them as interchangeable “healthy lighting metrics,” confusion becomes inevitable.

But if each metric is placed back into its proper role, many of the current ambiguities start to resolve themselves.

---

Why is traditional lighting language no longer sufficient?

In the past, the most familiar language of lighting was built around:

• illuminance
• luminance
• correlated color temperature (CCT)
• color rendering (CRI)
• light distribution
• glare

These metrics are still essential. They primarily serve visual tasks and spatial quality:

• Can we see clearly?
• Is it comfortable?
• Are colors accurate?
• Is the space bright enough?

---

But the scope of lighting has expanded

A growing body of research now shows that light also affects:

• circadian rhythms
• alertness
• emotional experience
• even certain behaviors and physiological responses

Standards and position statements from the International Commission on Illumination have clearly indicated that: the traditional photopic system is not sufficient to fully describe human responses related to ipRGCs (intrinsically photosensitive retinal ganglion cells).

---

This changes the fundamental questions

The industry can no longer stop at asking:

• “How many lux is this space?”
• “Is it 3000K or 4000K?”

Instead, it needs to ask:

• Which photoreceptive channels is this light stimulating?
• What does this imply for vision, circadian regulation, and emotional response?

---

This is the real context behind new metrics

This shift is precisely why metrics like:

• CAF
• CS
• EML
• EDI / DER

have emerged.

They are not just “new terminology,” but attempts to extend lighting language from: visual description → human biological interaction

In other words: from “how the space looks” to “how light actually affects people.”

The human eye doesn’t just “see” — it also “feels” light.

In the lighting industry, the most commonly discussed elements are rods and cones. That’s not wrong.

Rods are mainly associated with low-light (scotopic) vision. Cones are responsible for color, detail, central vision, and typical daytime visual functions.

But today we know that, beyond rods and cones, there is another critically important photoreceptive pathway in the human eye: ipRGCs (intrinsically photosensitive retinal ganglion cells).

These are associated with melanopsin, are more sensitive to short wavelengths, and are closely related to non-visual responses such as circadian rhythms, pupil response, and alertness.

However, to be more precise, what truly needs to be considered is not three systems, but five classes of photoreceptor channels: S-cone, M-cone, L-cone, Rod, and Melanopsin / ipRGC.

What CIE S 026:2018 establishes is a standardized metrology framework based exactly on these five photoreceptors.

In other words, for the first time, the industry has a shared language that is not only about “what can be seen,” but about “how light stimulates the five types of receptors.”

This is a critical shift. Because it means the lighting industry is moving from “spatial output” to “human input.”

---

What do CAF, CS, and EML actually represent?

1. CAF: closer to a “spectral efficiency ratio” mindset

CAF (Circadian Action Factor) has long been used to compare the potential of different spectra to stimulate circadian-related responses.

Its core logic is straightforward: under the same visual lighting conditions, is this spectrum more biased toward “circadian effect” or “visual effect”?

So CAF is essentially a weighted efficiency ratio. It helps compare different SPDs under the same photopic lux to determine which produces stronger circadian-related stimulation.

This approach is not without value. Its advantages are simplicity, intuitiveness, and suitability for early-stage comparisons.

But it also has clear limitations:

• First, it reflects spectral properties rather than actual human exposure dose.
• Second, it does not inherently include time, spatial context, viewing direction, or actual eye exposure.
• Third, it is not the primary shared language in current international standards.

So CAF helps you understand “how biased a spectrum is,” but it is not suitable as a complete coordinate for human response.

CAF is more like a spectral screening tool, not a full human-centric lighting language.

---

2. CS: closer to a “specific physiological response model”

CS (Circadian Stimulus) has also had significant influence in recent years, especially in North America. Its logic differs fundamentally from CAF.

Rather than being a simple ratio, CS attempts—through a circadian phototransduction model—to map spectral stimuli onto a response scale related to melatonin suppression.

UL’s DG 24480 also proposes design targets based on this type of framework. The strength of CS is that: It goes beyond saying “more or less biased,” and tries to quantify “how strong the circadian system stimulation is.”

But this is also where the challenge lies. Once a metric moves from “describing input” to “predicting response,” it inevitably introduces modeling assumptions:

• What spectral sensitivity functions are used
• How rod, cone, and melanopsin interactions are handled
• How dose-response is defined
• How exposure duration is treated
• How pupil state, timing, and exposure history are incorporated

As a result, CS has been accompanied by considerable methodological debate.

So a fair summary would be: CS is important, but it is better suited as an application-layer or response-layer model, rather than a foundational common coordinate system for the entire industry.

---

3. EML: closer to a “transitional language for application”

EML (Equivalent Melanopic Lux) has been widely promoted in application contexts such as WELL.

Its key contribution is that it helped many people realize, for the first time: Not all lux are the same.

From a communication and adoption standpoint, EML has played a significant role. It translates complex spectral–receptor relationships into a format that is easier to understand and specify in project requirements.

However, from a stricter standardization perspective, EML is not the ideal end state. The industry has increasingly shifted toward melanopic EDI, and further toward the more comprehensive α-opic EDI / DER framework, because these align better with the standardized structure defined in CIE S 026 and enable consistent use across organizations and systems.

So in one sentence: EML is a bridge toward human-centric lighting—but not the most suitable final coordinate system.

---

Why are EDI / DER closer to a true coordinate system?

Because they resemble a system that can be recorded, compared, and transmitted—like a “spectrum.”

1. EDI: describing “equivalent stimulus dose”

EDI (Equivalent Daylight Illuminance) can be understood as: How much illuminance from standard daylight (D65) would be required to produce the same level of stimulation for a given photoreceptor?

This allows results from different spectra to be compared within a unified framework.

---

2. DER: describing “stimulation efficiency”

DER (Daylight Efficacy Ratio) can be understood as: How efficient a given light is, per unit of photopic illuminance, at stimulating a specific photoreceptor.

CIE TN 015:2023 clearly defines the relationship: melanopic EDI = illuminance × melanopic DER

Together, these two quantities are powerful:

• EDI reflects the actual dose reaching the human body
• DER reflects the intrinsic efficiency of the light spectrum

One is exposure-focused, the other is source-focused. This combination is exactly what manufacturers, designers, control systems, and simulation tools need.

---

Why move from single m-EDI toward a full EDI / DER framework?

This is not a rejection of melanopic metrics. In fact, many recent consensus recommendations are indeed centered on melanopic EDI.

For example, Brown et al. (2022) suggest indoor light exposure guidelines such as:

• At least 250 lx during the day
• Preferably below 10 lx in the evening
• As close to 1 lx as possible at night

These are important. But looking further ahead: Humans do not respond to light through melanopsin alone.

Visual performance, color discrimination, adaptation, spatial perception, aspects of emotional experience, and more complex neural responses all involve the combined action of rods, S/M/L cones, and ipRGC pathways.

So if the industry aims to build a future-oriented human-centric lighting coordinate system, focusing only on m-EDI is not enough.

What we need is a more complete EDI / DER framework: Not just melanopic—but incorporating stimulation across all five photoreceptor classes into a unified language.

This does not mean every project must present all five values. It means: The foundational language of the industry should leave room for a complete human model.

---

From “selling light” to “describing humans”: the industry needs a new staff notation

I like to use an analogy: EDI / DER in human-centric lighting is like musical notation in music.

Musical notation is not the music itself, but it is the foundational language that allows music to be recorded, transmitted, reproduced, and collaboratively created.

EDI / DER is similar.

It is not sleep itself.
Not emotion.
Not comfort.
Not spatial aesthetics.

But it provides a way to more precisely describe: What this light is doing to the five photoreceptive channels of the human body.

With such a coordinate system, many long-standing ambiguities in the industry can finally be addressed collaboratively:

• LED manufacturers can provide more meaningful spectral data
• Luminaire manufacturers can define products in terms of human impact
• Control systems can move beyond brightness and CCT to modulating receptor stimulus
• Simulation tools can evolve from illuminance-based to human-input-based modeling
• Designers can move from “feels healthier” to “designing with coordinates”

Without such a system, the industry easily remains stuck in vague language:

More natural
Closer to daylight
More circadian-friendly
More comfortable
Healthier

These terms are not useless—but without an underlying framework, they struggle to become a shared language across organizations and product chains.

The real value of EDI / DER lies in this: For the first time, “how light affects humans” can be written down—like a score.

---

What does this mean for LEDs, luminaires, systems, and designers?

For LED and module manufacturers

Future competitive data cannot be limited to lm/W, CCT, and CRI.

SPD and α-opic / EDI / DER information will become increasingly critical.

---

For luminaire manufacturers

In the future, luminaires won’t just deliver lumens into space.

They will deliver specific receptor-stimulation structures to the human eye.

---

For control system manufacturers

Control strategies should no longer stop at “what time to switch to what CCT and what dimming level.”

A more advanced control objective should be: To achieve a target balance of stimulus dose and experience
for a given time, space, task, and user group.

---

For designers

Human-centric lighting design will go beyond “cooler in the morning, warmer in the evening.”

It will require thinking in terms of:

• Which photoreceptors this light primarily stimulates
• What the actual dose at the eye level is
• How to balance visual performance, circadian support, and emotional experience
• How daylight, electric light, reflections, and viewing direction interact

Once designers start thinking this way, lighting design evolves from “placing fixtures” to “modulating human response.”

---

Final point: the industry doesn’t lack terms—it lacks the ability to read the “score”

CAF, CS, EML, EDI / DER…
These terms often feel confusing not because they lack importance, but because they are frequently discussed at the same level.

In reality, they answer different questions:

• CAF → more like a spectral efficiency ratio
• CS → more like a specific physiological response model
• EML → more like an application-layer transitional language
• EDI / DER → closer to a standardized, computable, and transferable coordinate system

If the industry truly wants to move from “illuminating spaces” to “effectively influencing people,” the next step is not to invent yet another concept— but to learn how to read this system.

Illuminance tells us how bright it is. EDI / DER begins to tell us how light acts on humans. And that may well be the real starting point of the human-centric lighting era.

---

CTA

If your organization is exploring:

• How to upgrade LED or luminaire data from traditional photometric parameters to a language closer to human-centric lighting
• How to integrate EDI / DER into product definitions, control systems, or design simulations
• How to establish lighting evaluation methods that address circadian rhythms, visual performance, and emotional experience

You’re welcome to get in touch.

From Percent Flicker and Flicker Index to SVM, PstLM, and PAVM — rethinking the dynamic relationship between light and humans

For more than a decade, the LED lighting industry has commonly used the statement: “Our lights are flicker-free.”

But the real question is: What does “flicker-free” actually mean?

• Is it because camera sensors cannot capture visible banding?
• Is it because the human eye cannot perceive flicker?
• Is it because Percent Flicker is low?
• Is it because SVM is within limits?
• Or because PstLM passes compliance thresholds?

From the perspective of healthy lighting, human-centric lighting, and age-inclusive environments, the issue is far more complex than this.

Flicker should more precisely be understood within the framework of Temporal Light Modulation (TLM). It is not a single phenomenon, nor can it be defined by a single metric.

More importantly, flicker is not only a property of the light source itself. It is: the dynamic light exposure experienced by a person in a specific space, at a specific time, performing a specific activity, under a specific physiological and psychological state.

This is why next-generation healthy lighting cannot be limited to discussions of spectrum, illuminance, correlated color temperature, or color rendering index, nor even melanopic EDI / DER alone.

We must also incorporate the temporal quality of light into the system-level understanding of lighting.

---

1| Flicker is not a single phenomenon, but a set of phenomena

In the lighting industry, all temporal variations in light output are often broadly referred to as “flicker.” However, strictly speaking, variations in light output over time can lead to different types of human perceptual and physiological responses.

At minimum, these can be categorized into three major visual phenomena:

1. Direct Flicker: visible flicker

This is the most intuitive form of flicker. It refers to situations where the human eye directly perceives light as flashing, pulsing, or fluctuating.

This typically occurs under conditions such as:

• low frequency operation
• high modulation depth
• poor driver quality
• unstable dimming behavior

It most commonly leads to:

• visual discomfort
• eye strain
• attention disruption
• headaches
• adverse reactions in sensitive individuals

This layer is typically described using metrics such as Percent Flicker, Flicker Index, and PstLM / Mp.

2. Stroboscopic Effect: motion discontinuity perception

This is not perceived as flicker in the light itself, but rather as temporal distortion of moving objects. For example:

• fan blades appearing to stop or reverse
• hand movements appearing discontinuous
• rotating machinery appearing to change speed incorrectly
• motion trajectories appearing segmented

This is not only a comfort issue. In environments such as industrial facilities, healthcare, sports, kitchens, and laboratories, it can become a safety risk.

This layer is primarily described using SVM（频闪可见度指标）.

3. Phantom Array Effect: spatial-temporal image splitting

This is a historically underestimated phenomenon. It typically occurs during rapid eye movements (saccades). Instead of fixating steadily on a light source, humans constantly move, scan, shift gaze, and change focus.

In such conditions, certain lighting systems—especially:

• high-intensity point sources
• linear LED luminaires
• vehicle headlights
• stage lighting
• retail display lighting

may produce the perception of multiple separated light images or streaks. This is known as the Phantom Array Effect.

It is particularly common in:

• high-brightness point sources
• exposed LED systems
• linear luminaires
• automotive lighting
• entertainment lighting
• retail environments
• outdoor nighttime lighting
• high-contrast visual scenes

In the latest TLM framework, this effect is increasingly quantified using PAVM (Phantom Array Visibility Measure).

This leads to a critical realization: Humans are not static lux meters. Humans move, scan, turn, fatigue, and respond to light in individualized ways.

---

2 | What do common flicker metrics actually represent?

Let us break down the main indicators.

1. Percent Flicker (modulation depth)

Percent Flicker is the most intuitive metric. It describes the relative difference between the maximum and minimum light output within a cycle.

In simple terms: the deeper the fluctuation, the higher the Percent Flicker.

If light output drops close to zero during a cycle, Percent Flicker becomes high. If light output remains nearly constant, Percent Flicker is low.

Its advantages:

• simple
• intuitive
• easy to interpret

However, its limitations are significant. It does not tell us:

• frequency of modulation
• waveform shape
• whether it is sine wave or PWM
• duty cycle characteristics
• perceptual visibility
• stroboscopic risk
• phantom array potential
• fatigue or headache risk

Therefore, Percent Flicker only answers: How deep is the fluctuation?

It does NOT answer: Is this fluctuation harmful or perceptible to humans?

These are fundamentally different questions.

2. Flicker Index

Flicker Index extends beyond Percent Flicker by considering the area under the waveform over time, relative to the average level.

This makes it more sensitive to:

• waveform shape
• duty cycle behavior
• PWM characteristics
• temporal distribution of light output

For example, two systems with identical Percent Flicker (e.g., 50%) may have very different perceptual impacts depending on whether the waveform is smooth (sine-like) or sharp (square-like).

Flicker Index can better capture these differences. However, it still does not represent full human response.

It does not incorporate:

• frequency-dependent visual sensitivity
• task-dependent perception
• motion-based stroboscopic effects
• eye-movement-related phantom array effects
• individual sensitivity differences
• long-term fatigue or neurological response

Thus, Flicker Index helps describe waveform quality, but cannot be directly equated with health risk.

3. Frequency

Frequency is a fundamental parameter in all TLM analysis.

It answers: How many times per second does the light fluctuate?

Measured in Hz. However, frequency alone is insufficient for risk assessment.

Because at the same frequency (e.g., 1,000 Hz):

• low modulation depth may be harmless
• high PWM modulation may still cause issues
• stationary viewing may reduce perception
• rapid scanning of bright sources may induce phantom array effects
• sensitive individuals may react differently

Therefore, frequency must always be interpreted together with:

• modulation depth
• waveform type
• duty cycle
• dimming level
• viewing condition
• task type
• population sensitivity

In other words: Frequency defines the time scale, but not the full human risk profile.

4. PstLM / Mp: direct flicker perception metrics

PstLM is used to describe short-term visible flicker perception (direct flicker).

It helps answer: Can a typical observer perceive flicker under given conditions?

In newer frameworks such as TM-39, Mp is also used for similar evaluation of direct flicker. Its key value is that it shifts analysis from purely physical waveform characteristics to human perception-based assessment.

However, it has limitations. It does not fully capture:

• motion discontinuity perception
• phantom array effects during eye movement
• migraine triggering potential
• long-duration fatigue effects
• cognitive load impact
• neurological responses (EEG/fMRI-level effects)

Thus, PstLM / Mp is essential for direct flicker assessment, but not sufficient as a complete human discomfort model.

5. SVM: stroboscopic visibility measure

SVM is designed to describe the visibility of stroboscopic effects.

It answers: Will moving objects appear discontinuous under this lighting?

It is critical in:

• manufacturing
• machining
• rotating machinery environments
• laboratories
• kitchens
• sports facilities
• medical procedure areas
• logistics and warehousing
• fast hand-motion tasks

Its importance is not whether light flickers, but whether motion perception is distorted. In environments with rotating machinery, stroboscopic effects can lead to misinterpretation of equipment motion, creating safety hazards.

However, SVM is not a universal indicator. It does not fully predict:

• phantom array effects
• headaches or migraine triggers
• long-term visual fatigue
• neurological sensitivity responses
• non-visual physiological effects

Therefore, SVM is a key metric for dynamic visual safety, but not a comprehensive “overall comfort score.”

6. PAVM: Phantom Array Visibility Measure

PAVM is a relatively recent and important development.

It corresponds to the Phantom Array Effect, which refers to the perception of multiple separated light images or streaks when the human eye moves—such as during saccades, scanning, or head rotation.

This type of effect has historically been underrecognized, largely because conventional testing assumes a stationary observer.

However, in real-world conditions, humans are never static. We constantly:

• walk
• turn our heads
• scan environments
• shift attention between near and far objects
• move between screens and lighting environments
• visually browse objects in retail spaces
• observe exhibits while in motion
• scan vehicle headlights at night

Therefore, PAVM addresses a critical gap in traditional flicker evaluation.

It is particularly relevant for assessing:

• retail environments
• exhibition and museum spaces
• stage and entertainment lighting
• high-brightness point sources
• automotive lighting
• linear luminaires
• outdoor nighttime lighting
• educational and medical environments

The emergence of PAVM highlights a fundamental shift: Flicker evaluation can no longer assume static viewing conditions. It must account for human motion, eye movement, and real behavioral patterns.

---

3 | Core differences between key metrics

We can summarize the relationships between the main indicators as follows:

Metrics	Main Description	Layer Classification	Main Purpose	Maximum Limitations
频闪百分比	Depth of light wave fluctuations	Description of physical stimuli	Fast judgment of modulation degree	Does not consider frequency and human body perception
频闪指数	Waveform area and duty cycle	Description of physical stimuli	Determine waveform quality	Not a human response model
Frequency	Speed of light fluctuations	Time parameters	Determine time scale	Cannot judge risk individually
PstLM / Mp	Directly visible flicker	Visual response model	Direct Flicker	Does not cover SE / PAE / physiological discomfort
SVM	Jumping sensation of moving objects	Visual response model	Stroboscopic Effect	Not suitable for representing overall health risk
PAVM	Images separated by eye movement	Visual response model	Phantom Array Effect	Still mainly a visual model

Therefore, we can draw a key conclusion:

Percent Flicker, Flicker Index, and Frequency describe the physical stimulus itself;
PstLM / Mp, SVM, and PAVM describe different layers of visual response;
but none of them alone can fully define human discomfort or neurological response.

---

4 | Why we cannot simply say “flicker-free”

The term “flicker-free” is fundamentally too vague.

When a product claims to be “flicker-free,” it should at minimum clarify:

• Is Percent Flicker very low?
• Is Flicker Index very low?
• Does it comply with PstLM requirements?
• Does it comply with SVM requirements?
• Has PAVM been measured?
• Was it tested at 100% output, or also at 10%, 20%, and 50% dimming levels?
• Is this based on single-luminaire testing or full spatial system testing?
• What dimming method is used—sine-wave, DC, PWM, or quasi-square waveform?
• Were sensitive populations considered, such as children, elderly users, or individuals prone to migraines or neurological sensitivity?

If these questions are not addressed, then “flicker-free” is merely a marketing phrase.

A genuinely professional statement should instead specify: Under defined output levels, defined dimming conditions, and defined control methods, the luminaire exhibits the following values: Percent Flicker, Flicker Index, Frequency, PstLM / Mp, SVM, and PAVM.

Only then can performance be verified, compared, and meaningfully applied in healthy lighting design.

---

5 | Flicker must be understood within the framework of “human × space × time × activity”

One of the most common mistakes in healthy lighting is reducing complex human responses to a single metric.

In the past, lighting was simplified into 勒克斯. Later, it became CCT. Then 比显指.
Today, many people reduce healthy lighting to melanopic EDI.

But real-world conditions are not that simple.

Human-centric lighting must return to four dimensions: human × space × time × activity

Flicker should be understood in the same way.

1. Human: different people have different sensitivity to flicker

Not everyone responds to light in the same way. Under identical lighting conditions, some people may experience no issues, while others may report glare, irritation, headaches, or fatigue.

Groups that require special attention include:

Children and adolescents

They spend long hours in classrooms, tutoring centers, study desks, and screen-based environments.

Flicker may not always be explicitly reported, but it can manifest as:

• reduced attention
• reading fatigue
• visual discomfort
• irritability
• decreased learning efficiency

Classroom lighting should not be evaluated solely by illuminance levels. It must also consider whether light remains stable, especially under dimming conditions, and whether low TLM performance is maintained in real operation.

Office workers and sub-healthy populations

Many office workers already experience:

• dry eyes
• headaches
• sleep disruption
• neck and shoulder tension
• reduced attention
• visual fatigue

High temporal light modulation (TLM) in office environments may not be the sole cause, but it can act as an aggravating factor. This is particularly relevant in:

• open-plan offices
• smart lighting systems with adaptive dimming
• sensor-based control systems
• environments mixing multiple lighting brands and drivers

In such cases, system-level TLM management becomes essential.

Migraine-sensitive and neurologically sensitive individuals

These individuals may not only be sensitive to brightness, but also to:

• flicker
• high contrast
• bright point sources
• dynamic lighting changes
• phantom array effects
• visual noise

For them, lighting discomfort is often more pronounced and complex. Therefore, healthy lighting should not only be designed for the “average user,” but also for sensitive populations.

Elderly users

The issue for elderly populations is not simply “more light.” They generally require:

• stable illumination
• soft visual environments
• low glare
• low flicker
• safe dynamic perception
• non-disruptive nighttime lighting

In corridors, kitchens, staircases, bathrooms, and long-term care environments, both stroboscopic effects and phantom array effects can significantly impact safety and spatial perception.

Medical and long-term care populations

In healthcare environments, users are often in a vulnerable physiological and psychological state. They may experience:

• poor sleep
• anxiety
• fatigue
• pain
• increased sensitivity to environmental stimuli

In such contexts, light is not just illumination—it is part of the recovery environment.

Therefore, in addition to illuminance, glare, and spectrum, TLM must also be considered as part of lighting design in hospitals, wards, nursing stations, and rehabilitation spaces.

2. Space: single-luminaire compliance does not guarantee system compliance

Many flicker evaluations remain at the single-luminaire level. However, in real environments, people are not exposed to one lamp—they experience:

• multiple luminaires
• multiple angles
• multiple reflections
• multiple control circuits
• multiple drivers
• multiple dimming states
• multiple visual tasks

A single compliant luminaire does not guarantee a compliant space. For example:

• phase differences between luminaires
• beat frequency between different drivers
• waveform inconsistencies across dimming groups
• reflections amplifying high-brightness artifacts
• differences between eye-level exposure and desk-level measurements
• perceptual effects during walking or scanning (including PAE-related phenomena)

Therefore, future lighting acceptance testing should not rely solely on single-luminaire reports.

Instead, it should evaluate real spatial exposure, including:

• user eye-level conditions
• desktop plane conditions
• wall and surface reflections
• circulation paths
• lighting scenes and modes
• nighttime operation modes
• multi-luminaire simultaneous operation
• scene transitions

This marks the shift from “luminaire flicker” to “spatial TLM exposure.”

3. Time: flicker risk varies across operating conditions

Flicker should not be evaluated only at 100% output.

In practice, many LED systems exhibit their most problematic behavior under dimmed conditions, especially:

• 10% dimming
• 20% dimming
• low-level nighttime modes
• sensor-driven dimming systems
• scene switching events
• daylight harvesting adjustments
• standby modes
• emergency operation modes
• low-duty-cycle PWM control

This is why healthy lighting products must provide TLM performance data across multiple dimming levels—not just full output conditions.

Time also includes human biological timing. Morning, daytime, evening, nighttime, and pre-sleep conditions all require different lighting strategies. Nighttime healthy lighting should not be defined only by “low blue light.”

It should also ensure:

• low illuminance
• low glare
• low melanopic stimulation
• low TLM
• low visual stress

A truly sleep-friendly lighting environment is not just warm in color temperature—it is stable, low-stimulus, low-fluctuation, and low-disruption as a system.

4. Activity: different tasks require different flicker management

The same luminaire can present very different risk profiles depending on the activity. For example:

Reading

Key concerns:

• direct flicker
• eye fatigue
• attention interference
• long-term visual stability

Office work

Key concerns:

• interaction between display and lighting TLM
• long-duration visual load
• stability under dimming conditions
• risk of eye strain and headaches

Industrial environments

Key concerns:

• rotating machinery
• moving objects
• stroboscopic effects
• safety misperception risks

Healthcare environments

Key concerns:

• precision tasks
• patient sensitivity
• long working hours and fatigue
• nighttime lighting in care stations

Retail and exhibition spaces

Key concerns:

• high-brightness point sources
• rapid visual scanning
• phantom array effects
• balance between visual comfort and product presentation

Residential and hospitality environments

Key concerns:

• emotional comfort
• relaxation
• sleep preparation
• nighttime safety
• low visual stimulation

Therefore, healthy lighting should not be defined as “one scenario = one metric set,” but rather as a system where different activities require different lighting recipes.

---

6 | The real meaning for all-age and sub-healthy populations

All-age healthy lighting is not about making light brighter or warmer.

Its real purpose is to address: the integrated lighting needs of different ages, physiological states, spaces, times, and activities.

Within this framework, flicker management represents the temporal quality of light. If we define lighting functions as:

• illuminance → visibility (“can I see clearly?”)
• CCT → visual atmosphere (“what does the light feel like?”)
• CRI → color accuracy (“are colors rendered correctly?”)
• EDI / DER → spectral biological stimulus (“how does the spectrum interact with physiology?”)

Then TLM management addresses something different: whether light is stable, non-disruptive, and temporally compatible with human perception and physiology.

This is especially important for sub-healthy populations.

Because sub-health is rarely a single condition—it is a cluster of chronic stress factors, 例如

• poor sleep
• eye fatigue
• mental tension
• headaches
• anxiety
• reduced attention
• heightened environmental sensitivity

In such states, every environmental stimulus can become a load.

Flicker, glare, high-intensity point sources, incorrect color temperature, and excessive nighttime stimulation can all accumulate into physiological stress.

Therefore, future healthy lighting should not only aim to be “bright” or “visually appealing,” but should instead focus on: a measurable, manageable, and verifiable low-burden lighting environment.

---

7 | What product development should look like

Future healthy lighting products should not only specify:

• power
• luminous flux
• color temperature
• 比显指
• beam angle
• UGR

They should also clearly provide:

• 频闪百分比
• 频闪指数
• Frequency
• PstLM / Mp
• SVM
• PAVM
• waveform characteristics
• duty cycle
• dimming curve behavior
• driver ripple characteristics
• TLM performance at 100%, 50%, 20%, and 10% output levels
• performance under different control protocols

Especially for products claiming:

• 健康照明
• eye protection lighting
• learning environments
• sleep-friendly lighting

Full transparency of temporal behavior is essential. Because “flicker-free” is not a scientific endpoint.

A truly professional product should instead state: Under defined output conditions, control strategies, and application scenarios, the TLM-related risks are controlled within specified ranges.

---

8 | What lighting design practice should evolve into

Lighting design should not rely solely on illuminance calculations or photometric files.

A more complete data structure is required: Photometry + Spectrum + Alpha-opic + TLM Profile

A healthy lighting design dataset should include:

• photometric distribution data
• spectral power distribution (SPD)
• EDI / DER metrics
• CCT / Duv
• CRI / TM-30
• 频闪百分比
• 频闪指数
• Frequency
• PstLM / Mp
• SVM
• PAVM
• dimming states
• control profiles
• scene tags

Only with this level of integration can we begin to build a true digital twin of healthy lighting.

Not just simulating how bright a space is, but simulating: how a human actually experiences dynamic light exposure over time, across activities and spatial contexts.

---

9. How site acceptance testing should evolve

Commissioning should no longer be limited to single luminaires or desktop illuminance measurements.

Especially in schools, offices, healthcare facilities, long-term care environments, hotels, high-end residential projects, and industrial spaces, evaluation should progressively include:

• single-luminaire TLM
• multi-luminaire combined TLM
• eye-level user exposure TLM
• task plane TLM
• surface reflection TLM
• dimming-state TLM
• nighttime-mode TLM
• scene-transition TLM
• PAE risk during movement and scanning
• SVM risk under dynamic mechanical operation

This marks the transition from: “product compliance” → “spatial compliance”

It is a critical step in moving healthy lighting from concept to engineering verification.

---

10 | Next-generation healthy lighting: beyond spectrum, beyond flicker

The lighting industry has gone through several stages of evolution:

1. from “having light” → “enough brightness”
2. from “brightness” → “energy efficiency”
3. from “efficiency” → “light quality”
4. from “light quality” → “healthy lighting”

But today, healthy lighting cannot be reduced to isolated indicators. We cannot only talk about:

• high CRI
• low blue light
• no flicker
• high EDI
• full spectrum
• eye protection
• sleep support

These are all meaningful, but incomplete. The next step is to return lighting to real human life:

• What kind of person?
• In what kind of space?
• At what time?
• Performing what activity?
• In what physiological and psychological state?
• Receiving what spectral, spatial, temporal, and dynamic light exposure?

This is the true foundation of human-centric lighting.

---

Conclusion: stop asking only “is there flicker?”

We should stop asking: Does this light have flicker?

And instead begin asking: Under what frequency, waveform, dimming condition, spatial configuration, task context, and population sensitivity does this system produce direct flicker, stroboscopic effects, phantom array effects, and potential visual or physiological discomfort?

This is not overcomplicating the problem. It is clarifying it. Because humans are not static instruments. Spaces are not single-luminaire test setups. Time is not an average value. And activities are not abstract scenarios.

True healthy lighting must shift from static parameters to dynamic exposure:

• from spectrum → time
• from single luminaire → spatial system
• from average user → real human diversity
• from product metrics → lived experience

This is the direction flicker research must evolve toward.

And it is the challenge that next-generation healthy lighting, all-age lighting environments, and sub-healthy-friendly spaces must address.

---

Lawrence Industry Observation

I believe the healthy lighting industry is approaching a major inflection point.

One group of companies will continue to rely on marketing language, 例如

• flicker-free
• eye-friendly
• full-spectrum
• low blue light

Another group will move into engineering language, including:

• SPD
• EDI
• DER
• PstLM
• SVM
• PAVM
• dimming profiles
• spatial exposure models

And one step further, the real leaders will operate in a human-centric language framework: human × space × time × activity × individual sensitivity

This is the direction I have consistently emphasized. Healthy lighting is not about optimizing a single metric in isolation.

It is about building a lighting environment system that is:

• measurable
• verifiable
• manageable
• continuously optimizable

Within this system, flicker management is not optional. It is a fundamental component.

2025 台湾照明环境奖获奖项目｜现场测量点评

美感要有，数据也要有；故事要讲，现场更要经得起测量。

一个照明奖项作品，真正值得被讨论的，不只是它的夜景好不好看，也不只是它拍起来是否漂亮。 更重要的是：它是否真的让人行走安全、视觉舒适、方向清晰、眩光可控，并且能够经得起现场测量。

这次测量的是2025台湾照明环境奖获奖项目：南港立体连通平台。

这是一个非常值得讨论的案例，因为它并不是单纯把桥体“打亮”，而是将一座城市交通连通设施，通过照明转化为一个夜间可被阅读、可通行，并且具备城市地标感的公共空间。

---

一、这个项目的核心价值：不是更亮，而是更可控

很多公共照明，尤其是人行天桥、连通平台、地下通道、车站周边空间，常见做法是：灯具越多越安心，照度越高越安全。

但现场经验往往恰恰相反。过亮可能造成刺眼；光源裸露可能引发眩光；明暗差异过大反而会影响视觉适应；地面看似很亮，但人眼实际感知到的却是一片混乱的亮度分布。

南港立体连通平台这个项目较为成熟的地方在于，它并没有单纯依靠大量向下照明去“灌亮”地面，而是通过结构本身、栏板、立面反射、光源遮蔽以及线性引导，建立起一种更柔和、连续且可控的夜间空间感。

这一点非常关键。因为人在夜间步行时，需要的并不是单一测点上的高照度（lux），而是整体视野中稳定的亮度秩序。

---

二、现场测量数据：低照度，但空间感成立

从 In.Licht Ultra 现场测量数据来看，其中一组具有代表性的数值如下：

指标	现场数据	初步解读
照度	12.1 lux	不高，但若分布均匀、眩光受控，对夜间通行是可成立
CCT 色温	3911K	接近 4000K，中性偏暖，适合交通连通空间
CRI / Ra	88.1	整体显色不错
R9	43	红色还原普通，非主要缺点，但有提升空间
x / y	0.389 / 0.396	色度落点稳定，接近中性白光区域
Foot-candle	1.12 fc	约等同 12.1 lux

另一组测点则显示：

指标	现场数据
照度	17.0 lux
CCT 色温	3992.7K
CRI / Ra	92.4
R9	64

这说明不同位置之间仍然存在差异，但整体趋势是一致的：它并不是高照度型设计，而是以低到中低照度为主，更强调光的分布逻辑与空间反射关系的设计。

12–17 lux 如果放在办公、阅读或商业展示场景中，当然远远不够；但如果放在夜间户外人行连通平台，就不能简单用室内照明标准来判断。

这类空间更重要的是：路径是否清晰？边界是否可辨识？台阶、转角、栏杆、出入口是否能够被看见？是否存在刺眼的光源？从暗处走向亮处、从亮处走向暗处时，眼睛是否能够顺利适应？

从画面观察，这个项目中的栏板、桥体结构与地面边界被连续地标示出来，光并不只是落在地面上，而是让人能够看见空间的“轮廓”。这正是它比一般公共照明更值得讨论的地方。

---

三、色温：约 4000K，清晰但不冷硬

现场 CCT 约在 3911K–3993K 之间，接近 4000K。这是一个对交通型公共空间而言相对合理的选择。

它不像 3000K 那样容易形成过度休闲或偏黄的感受，也没有 5000K 以上冷白光常见的刺眼与紧张感。对于车站连通、城市换乘、人行平台这类需要方向感与辨识度的空间，约 4000K 的色温能够维持清晰、干净且具公共属性的视觉感受。

但这并不意味着 4000K 永远是标准答案。如果场地邻近住宅、酒店、河岸或休憩广场，或者使用时间进入深夜阶段，仍然可以考虑分区或分时段控制。例如在晚间高峰维持清晰的导向照明，而在深夜降低整体亮度或局部调暖，从而在公共安全与夜间环境干扰之间取得更好的平衡。

这也是未来城市照明应当发展的方向：不再是用单一固定色温贯穿全天，而是根据人、空间、时间与活动目的进行动态调整。

---

四、显色：整体表现良好，但红色还原仍有提升空间

从 CRI 与 TM-30 的角度来看，这个项目的显色表现属于可接受到良好的水平。

指标	数值
CRI / Ra	88.1
Re	82.7
R9	43
TM-30 Rf	86.9
TM-30 Rg	96.6

另一组测点则为：

指标	数值
CRI / Ra	92.4
Re	88.8
R9	64
TM-30 Rf	88.3
TM-30 Rg	96.5

第一，Ra 约 88–92，代表整体色彩还原能力不差。对于公共通行空间而言，这一水平已经明显优于许多只追求成本与效率的户外照明方案。

第二，TM-30 的 Rf 约 86.9–88.3，说明色彩保真度较为稳定；Rg 约 96.5–96.6，代表色彩饱和度没有被刻意放大，因此整体视觉呈现相对克制、干净，不会出现过度艳丽或不自然的色彩效果。

第三，R9 约 43–64，则显示红色还原能力处于一般到尚可的区间。对于一般行走空间而言，这并不是关键性问题；但如果该区域需要更好的面部识别、公共艺术呈现、植栽夜景效果、商业街区活力，或更高质量的城市夜间界面，那么 R9 仍有提升空间。

这也提醒我们：CRI 并不是全部，Ra 高并不代表所有颜色表现都良好。未来的公共照明评估，应当逐步将 TM-30、R9、光谱分布以及实际场景需求一并纳入，而不只是依赖单一的 Ra 数值。

五、节律与夜间健康：低 m-EDI 是合理的

这次测量也包含与人本照明相关的 HCL 指标：

指标	数值
CAF	0.55
EML	8.07
m-EDI	7.32 lux
S/P Ratio	1.68
CS	0.01

另一组测点则显示：

指标	数值
CAF	0.60
EML	12.45
m-EDI	11.29 lux
S/P Ratio	1.78
CS	0.02

这些数据说明，它对夜间生理节律的刺激很低。对于夜间户外公共通行空间而言，这反而是合理的。

白天的光应当帮助人保持清醒、同步昼夜节律并提升警觉性；夜间的光则应当保证安全通行，同时尽量减少对睡眠以及周边夜间环境的干扰。

所以这里的重点并不是把 m-EDI 做高，而是要确认：

• 有足够的视觉安全
• 没有过度眩光
• 没有过度蓝光刺激
• 没有把夜间公共空间做成“白天化”

这个项目在这一点上是比较克制的。它让人看得见，但并没有试图把夜晚变成白天。这一点是值得肯定的。

---

六、频闪：SVM 可控，但仍不应只依赖单一数值

频闪数据方面，补充画面显示：

指标	数值
SVM	0.44
频闪百分比	99.9%
频闪指数	0.50
Frequency	17,910.49 Hz

另一组测点则为：

指标	数值
SVM	0.27
频闪百分比	99.9%
频闪指数	0.33
Frequency	13,930.38 Hz

这组数据很适合用来做科普。很多人看到 percent flicker 99.9，第一反应会是：是不是很严重？但不能这样简单理解。

频闪风险不能只看 percent flicker，还要结合频率、波形、调制深度、SVM、PstLM，以及人的活动情境一并判断。

这里 SVM 为 0.27–0.44，代表可见频闪风险相对可控；同时频率达到约 14–18 kHz，已远高于一般人眼可感知的闪烁范围。因此，即使 percent flicker 数值较高，也并不意味着普通使用者一定会感受到明显闪烁。

但这并不意味着可以完全忽略这一问题。因为在公共空间中，往往存在多种使用情境：手机拍摄、车辆行驶视角、快速移动状态、边走边看的人群、老年人、儿童、对闪烁敏感的人，以及在不同调光状态下驱动性能的变化。

因此，更严谨的说法应当是：
从 SVM 与高频率来看，可见频闪风险相对可控；但 percent flicker 与 flicker index 偏高，因此仍建议在不同调光输出条件以及不同灯具位置下进行更完整的复测与验证。

这也说明了现场测量的重要性。仅仅查看灯具规格书，很难了解在实际安装之后，受驱动器、调光系统、配线以及控制条件影响时，最终在时间域上会呈现出怎样的光输出特性。

---

七、真正的优点：它用“光的秩序”取代“光的堆叠”

这个项目最值得肯定的，不是某一个单一指标非常漂亮，而是整体策略相对成熟。

它没有把公共安全简化为“越亮越好”；没有把桥体照明做成刺眼的景观灯秀；没有让光源直接暴露在主要视线范围内；也没有让人行空间充满凌乱的明暗斑块。

它做的是：

• 通过栏板照明建立行走边界
• 通过结构照明塑造桥体节奏
• 利用反射光减少直接眩光
• 用低照度维持夜间尺度
• 用中性色温保持公共空间的清晰感
• 以相对克制的光输出降低环境干扰

这是更高阶的公共照明逻辑：不是把灯装满，而是让光出现在该出现的位置；不是让地面最亮，而是让人的视觉环境更稳定；不是追求一张好看的夜景照片，而是让使用者在夜间通行时，感到安全、自然且不被打扰。

---

八、仍可补强的专业验证项目

当然，仅凭一组或几组现场截图，还不足以完整定义整个项目的照明品质。如果要做更完整的获奖项目评估，建议至少补充以下几类数据：

1. 多点位照度与均匀度 : 包括入口、出口、桥中央、转角、楼梯、扶梯连接处、栏板旁、玻璃反射区域、桥下阴影区域等。
2. Vertical illuminance at eye level: 人们在空间中真实感受到的光，并不只是地面水平照度。眼位垂直照度更能反映人脸识别、视觉安全、空间感以及夜间舒适度。
3. 亮度分布与眩光风险: 尤其需要观察光源是否进入主要视线方向、玻璃栏板是否产生反射眩光，以及车道侧是否存在视觉干扰。
4. 不同天气条件: 雨天、湿滑地面、雾气以及玻璃反光都会改变夜间的视觉感受。台北气候潮湿多雨，因此这一点尤为重要。
5. 不同时段控制: 傍晚、通勤高峰、深夜、末班车后，不同时段的照明任务并不相同。未来应更积极导入分时段调光与人流感测机制。
6. 完整频闪复测: 包括不同灯具类型、不同调光比例、不同供电回路，并同时记录 SVM、PstLM、percent flicker、flicker index 以及频率。

当这些数据补充完整后，这个项目就不再只是一个“好看”的获奖作品，而是有潜力成为台湾公共照明迈向“可测量、可验证、可优化”的重要参考案例。

---

九、给照明行业的提醒：奖项应当进入“可验证时代”

这次测量最大的意义，不只是评价南港立体连通平台，而是提醒整个照明行业：

照明奖项不能只停留在照片、叙事与设计概念。

未来的好照明，应该至少同时回答几个问题：

• 现场是否真的舒适？
• 眩光是否被控制？
• 照度是否足够但不过量？
• 光谱是否符合场景与时间？
• 显色是否支撑使用需求？
• 频闪是否被检查？
• 夜间是否减少对人与环境的干扰？使用一段时间后，是否仍维持原设计品质？

这些都不能只靠感觉，也不能只靠设计图，更不能只靠灯具规格书。必须回到现场，用仪器测量，用人的感受验证，用场景需求校准，用时间维度追踪。这才是公共照明真正应该走向的下一步。

---

十、结语：公共照明的成熟，不是更亮，而是更懂人

南港立体连通平台这个项目值得肯定。它并不是一个完美答案，但它提出了一个很好的方向：公共照明不必依靠高照度取胜。

真正成熟的城市照明，是在安全、舒适、节能、低眩光、低干扰与可验证性之间取得平衡。

这个项目的价值，不只是“好看”，而是它让我们看到：

• 光可以服务行走
• 光可以组织空间
• 光可以降低眩光
• 光可以尊重夜晚
• 光也可以被数据重新理解

南港立体连通平台的价值，不在于它让城市变得更亮，而在于它让光的使用更有秩序。

而这，正是未来公共照明最需要的能力。

美感要有，数据也要有；故事要讲，现场更要经得起测量。

From colored-light products to “emotional navigation systems”—how should the lighting industry evolve?

Abstract

“Emotional lighting” has become a popular topic in recent years, yet most products still remain at the stage of color tuning, atmosphere creation, and storytelling. They are still far from truly influencing human states. Effective emotional lighting should not merely output colored light; it should evolve into a measurable, verifiable, feedback-driven, and iterative closed-loop system integrating light × human × brain.

This article explores the underlying logic of emotional lighting from perspectives such as circadian rhythm, alertness, the limbic system, reward mechanisms, EDI/DER, and PBM. It also proposes a more meaningful direction for the industry: future lighting products should not just sell luminaires, but move toward adaptive emotional navigation systems.

---

Introduction

In recent years, concepts like “pastel lighting,” “emotional lighting,” “ambient lighting,” “color therapy,” and “dopamine spaces” have gained significant popularity.

However, if we are honest, most so-called emotional lighting products in the industry are still far from truly influencing human states.

Many products are highly capable of changing colors and telling compelling stories; yet they often do not know:

• Whether the user is more relaxed or more irritated at that moment
• More alert or more fatigued
• Comforted or overstimulated

In other words, most “emotional lighting” today is essentially lighting with visual styling, rather than evidence-based human-centric systems with a closed-loop validation process.

And this is precisely the gap the industry must address next.

---

1. Why most “emotional lighting” today is not yet professional

The conclusion first: If emotional lighting is to produce real effects, it cannot remain at colored light output—it must evolve into a stimulus–perception–feedback–adjustment system.

Because “emotion” has never been something controlled by a single color button.

It involves at least four layers:

• Layer 1: Visual perception
  A space that appears warmer, softer, lighter, or more dramatic will certainly influence subjective preference.
• Layer 2: Physiological arousal
  The same light can make a person more alert—or more fatigued; more focused—or more irritable.
• Layer 3: Circadian effects
  The key factors are not just color, but also timing, duration, dose, direction, and prior light exposure history.
  The CIE’s 2024 position statement reiterates that “the right light at the right time” should be characterized using the CIE S 026 α-opic framework, rather than relying solely on CCT or traditional illuminance.
• Layer 4: Stable emotional experience
  Feelings such as happiness, relaxation, safety, and healing are rarely caused by a single color. They are shaped by multiple factors, including sleep quality, alertness level, stress state, circadian alignment, reward system activity, and environmental security.

Therefore, being able to adjust RGB does not mean one understands emotional lighting.

---

2. Why colored lighting often “feels right” but isn’t necessarily effective

Because most colored-light products control output, but not dose. Common industry practices include:

• Preset scenes (happy orange, healing blue, meditation purple, energetic red)
• App-based interaction
• Pairing with music, scent, or marketing narratives
• Assuming users will “feel better”

The core issue is not aesthetics—it is the lack of measurement and validation.

If an emotional lighting system does not know the user’s:

• Current alertness
• Stress level
• Fatigue
• Sleep condition
• Sensitivity to stimuli
• Emotional stability

…it cannot determine what light should be delivered, nor whether the intended effect has been achieved.

Research on light interaction with the brain’s limbic system suggests a valuable direction. Instead of relying only on subjective questionnaires, it explores extracting brain-state indicators—such as anxiety tendency, depression tendency, tension level, sleep index, brain fatigue, external/internal focus, and hemispheric dominance—through frontal EEG, algorithms, and database modeling.

These metrics may not yet constitute industry standards, but they point to a critical shift: If emotional lighting is to advance, devices must evolve from “emitting light” to “measuring humans.”

---

3. How light actually influences emotion (beyond “dopamine”)

A common claim today: “This light stimulates dopamine and makes people happy.”

This is overly simplistic. A more rigorous statement is: Light does not directly “create happiness.” Instead, it alters the emotional baseline through:

• Retinal input
• Circadian synchronization
• Sleep homeostasis
• Activation of alertness systems
• Indirect modulation of brain regions related to emotion and reward

Over the past two decades, research has shown:

• Nighttime light, especially short-wavelength stimuli, significantly suppresses melatonin
• The spectral sensitivity of this effect differs from that of the visual system, indicating non-visual photoreception beyond rods and cones
• Melanopic EDI is a strong predictor of nighttime melatonin suppression

In other words, the foundation of emotional lighting is not merely “color psychology,” but a continuous system: Light → Eye → Brain → Circadian Rhythm → Emotion

---

4. Three terms the industry should stop conflating

To avoid conceptual confusion:

• Orexin – regulates wakefulness, motivation, and goal-directed arousal
• Serotonin (5-HT) – associated with mood stability and daytime state
• Dopamine – linked to reward prediction, novelty, and motivation

If emotional lighting is to move toward interdisciplinary collaboration, these terms must be used precisely from the outset.

---

5. Don’t treat dopamine as an ever-increasing “happiness knob”

Terms like “dopamine lighting” or “dopamine spaces” are fine as marketing labels—but not as scientific models.

Dopamine is better understood as a signal of:

• Reward prediction error
• Novelty
• Motivation and exploratory behavior

When stimuli exceed expectations, dopamine responses increase. But once stimuli become predictable and repetitive, the effect diminishes and may even fade into the background. This has direct implications:

• Short-term stimulation ≠ long-term value
• Feeling “excited” ≠ being sustainably “happier”

Mature emotional lighting should prioritize:

• Circadian alignment
• Daytime activation
• Minimal nighttime disruption
• Sleep recovery
• Stress regulation
• Emotional stability
• Resilience

---

6. Light and emotion involve more than dopamine

Light’s influence extends beyond the reward system:

• Orexin → wakefulness, motivation, goal-oriented activation
• Serotonin → mood stability, daytime function, seasonal mood variation

A classic Lancet study showed that serotonin production in the brain correlates positively with daily light exposure. Thus, a more evidence-based statement is not: “This light makes you release happiness hormones.”

But rather: Appropriate light exposure—at the right time, dose, direction, and spectrum—supports a better emotional baseline through circadian, alertness, sleep, and neural pathways.

Light does not directly create happiness; it creates the conditions under which happiness, stability, recovery, and focus are more likely to occur.

---

7. Why the limbic system matters for lighting

Emotion does not occur in luminaires or color palettes—it occurs in the brain.

Recent research combining fMRI and EEG has explored how different lighting conditions affect emotional brain regions, proposing a Limbic System Score (LSS) to quantify interactions between parameters such as CCT, CRI, flicker, illuminance variation, and exposure time.

Key observations include:

• Excitement peaks around 3000–4000K
• Happiness peaks around 4000K
• Higher CCT (~5700K) tends to suppress emotional activation

Additional findings suggest relationships between specific brain regions and CCT thresholds, for example:

• Calcarine → minimal negative emotion response around 4000K
• Frontal Superior → strongest emotional excitation around 4400K
• IFG & MCC → highest stability around 4200K

These are not yet universal standards, but they demonstrate: Emotional lighting is measurable—not purely subjective.

More importantly, this work shifts emotional experience from narrative into a quantifiable, modelable, and iterative “light recipe language.”

---

8. From EDI / DER to PBM: building a true “stimulus language”

For emotional lighting to become verifiable and scalable, the industry must move beyond vague descriptors like “warmer” or “softer.”

Three key concepts:

1. EDI (Equivalent Daylight Illuminance)
   Describes effective stimulus at the eye.
   The key question is not just “how bright,” but how much effective stimulus the eye receives.
2. DER (Daylight Efficacy Ratio)
   Compares non-visual effectiveness under equal visual brightness, enabling cross-product comparison.
3. PBM (Photobiomodulation)
   Requires full parameterization:
   • Wavelength
   • Intensity
   • Energy density
   • Exposure time
   • Pulse structure
   • Thresholds
   • Safety limits

Effective light is not about “looking right”—it is about reaching the correct dose window.

---

9. PBM’s key lesson: every stimulus has thresholds

• Too little → ineffective
• Too much → potentially inhibitory

Thus, future emotional lighting must incorporate:

• Threshold definition
• Dose modeling
• Safety boundaries
• Closed-loop validation

---

10. Emotion is not switching—it is navigation

Lighting should not force a direct jump from “nervous” to “happy.” A more realistic pathway:

• nervous → neutral
• neutral → relaxed
• relaxed → happy or alert

This reflects actual human regulatory processes. Therefore, advanced emotional lighting is not a fixed scene library—it is a path-planning system.

It must answer:

• Where is the user now?
• What is the target state?
• What path is appropriate?
• How long should transitions take?
• What stimulus intensity is optimal?
• When should stimulation be reduced or increased?
• When should intervention be avoided entirely?

---

11. The inevitable future: adaptive emotional navigation systems

Next-generation systems will:

• Continuously assess user state
• Dynamically adjust lighting strategies
• Optimize transitions over time

Like navigation systems:

• Determine current position
• Select the optimal route
• Adjust in real time

Not forcing the same “shortest path” every time.

---

12. Beyond luminaires: defining a system-level architecture

Future value lies not in hardware alone, but in integrating:

• EDI / DER
• Temporal programming
• Spatial distribution
• Directional control
• Stimulus dosing
• State sensing
• Feedback correction
• Path optimization

Lighting becomes: A programmable, measurable, verifiable, and navigable human-centric system—not just a product.

---

Conclusion

The next step in emotional lighting is not better color control—it is deeper human understanding.

Color is not wrong. Atmosphere is not wrong. “Dopamine spaces” and “healing light” are not wrong. But stopping at color imagination is insufficient.

The industry must move toward real research on: Light → Human → Brain → Emotion

Both healthy lighting and emotional lighting must converge toward:

• Measurement
• Understanding
• Validation
• Human-state optimization

Ultimately, emotional lighting should not simply “paint spaces with trendy colors.” It should function like music:

• With rhythm
• Dynamics
• tonality
• and a definable symbolic system
• EDI / DER define what stimulus reaches the eye
• PBM defines thresholds and dosage
• Adaptive emotional navigation systems define how lighting regulates human states

At that point, lighting is no longer about selling colored light—It becomes a new paradigm of: devices, algorithms, and evidence systems that redefine how humans interact with light.

---

Postscript

If the industry is willing to seriously advance this field, I strongly recommend initiating a round of cross-disciplinary collaborative research, involving:

• Lighting companies
• Sensor technology firms
• Sleep medicine specialists
• Psychologists
• Neuroscientists
• Spatial designers
• Scenario/experience operators

Because the real barrier in emotional lighting is no longer luminaire development.

It is this: Are we willing to acknowledge that future lighting products must increasingly resemble human-centric technology systems?

---

Scientific Basis / Further Reading

• CIE, Position Statement on Integrative Lighting — Recommending Proper Light at the Proper Time, 3rd ed. (2024)
• CIE S 026:2018, System for Metrology of Optical Radiation for ipRGC-Influenced Responses to Light
• ISO/CIE TR 21783:2022, Integrative lighting — Non-visual effects
• Brown et al., PLOS Biology (2022), recommendations for indoor daytime/evening/night light exposure
• Brainard et al., Journal of Neuroscience (2001), action spectrum for human melatonin suppression
• Thapan et al., Journal of Physiology (2001), melatonin suppression and non-visual photoreception
• Lambert et al., The Lancet (2002), sunlight and brain serotonin turnover
• Korshunov et al., Frontiers (2017), dopamine and circadian regulation
• Huang et al., Dose-Response (2009 / 2011), biphasic dose response in PBM/LLLT
• de Freitas & Hamblin, Frontiers in Neuroscience (2016), review of PBM mechanisms

从HP Opto、Philips到PE与地缘政治：他曾定义一个时代，最终却成为时代转身中的待售资产

导语

如果今天再回头看 Lumileds，我认为它早已不是一家普通公司的兴衰故事。

它更像是一面镜子。映照的不是单一企业的成败，而是过去三十多年里，LED产业的主导权如何从美国光电实验室、欧洲品牌体系，逐步转向亚洲制造体系、资本市场与地缘政治框架。

这家公司出身 HP Inc. 光电血脉，曾经推动红光LED、高亮度AlInGaP、LUXEON高功率封装、车用LED等多条技术路线，也实实在在支撑了 Philips Lighting 从传统光源向LED的战略转身；但它自己，却在产业成熟、资本改写与地缘政治升温的洪流中，被一再转手、反复估值，最终两度因美国国安疑虑而交易告吹。它的源流可以一路追溯到HP在1960年代末期推进LED商业化，之后经 Agilent Technologies 演化，最终在1999年形成 Lumileds。

而我对这段历史，不只是旁观。1998年，我曾造访槟城的HP Opto工厂。后来在东贝光电担任采购时，东贝曾是Lumileds在台湾最大的客户之一；再之后，我又参与十城万盏路灯项目，采用了Lumileds的产品，与他们有过很多互动。也因此，Lumileds在我眼中，从来不只是新闻稿上的一家外商公司，而是一个我曾在不同年代、不同位置上近距离接触过的产业角色。

一、Lumileds的根，其实不在照明，而在HP Inc.

很多后来进入行业的人，容易把Lumileds直接理解成Philips旗下的LED公司。这样说不能算错，但远远不够。

Lumileds的根，真正埋在HP Inc.的光电半导体体系里。

早在1968年，HP Inc.就已开始推进LED商业化；1970到1972年间，HP的LED已被用在HP-35计算机与数字手表。再往后，这条技术线一路延伸到高亮度AlInGaP红黄光LED，之后经由Agilent Technologies与Philips的合资，于1999年形成Lumileds。换句话说，Lumileds的血统，不是从传统照明公司里长出来的，而是从光电半导体工业里长出来的。

这件事很重要。因为它决定了Lumileds与很多传统照明企业本质上的不同。它不是先懂灯具、通路、工程与品牌，再慢慢学会LED；它是先懂材料、外延、晶片、封装、可靠性，再一步一步走向应用与照明。

这种出身，决定了Lumileds在LED产业早期能扮演的角色，远远不只是供应商，而更像是一个技术平台。

二、HP Inc. Opto不只是前身，更是Lumileds产业早期的人才母体

如果只把 HP Opto 看作 Lumileds 的前身，其实还是低估了它的历史地位。

在我看来，HP Opto 更像是 LED 产业早期的一所“实战学校”。

它培养出的，不只是工程师，也包括懂材料、懂制程、懂封装、懂可靠性、懂产品化、懂全球运营的一整代专业人士。像 Michael Krames 这样后来在 LED 产业极具代表性的技术领袖，就是沿着 HP optoelectronics 到 Philips Lumileds 这条路径成长起来的。

从整个行业后续的人才流动来看，这个体系外溢出去的人才，也陆续流向 Lumileds、Cree、Bridgelux，以及欧美与中国许多 LED 与照明企业，成为不同公司中的技术骨干、管理层与关键专业人士。这部分未必能用一份单一名单完整穷尽，但作为长期身处行业的人，这个脉络其实非常清楚。HP 到 Lumileds 这条线，不只输出了产品，也输出了一整套方法论：如何把实验室里的光电技术，做成可以量产、可以验证、可以打进全球市场的产品与组织。这条主线对整个 LED 产业的外溢影响极深。

所以，如果说 Lumileds 是 Philips LED 转型的技术引擎，那么 HP Opto 更像是整个 LED 时代早期的人才母体，甚至可以说，是半导体照明产业的一所“黄埔军校”。

三、它曾经不只是领先，而是定义过 LED 的一個時代

Lumileds 最值得敬重的地方，不只是曾经很强，而是它确实定义过几个市场。

从 1990 年代中后期到 2000 年代初，Lumileds 接连推出了 SuperFlux、SnapLED、第一代高功率 LED、LUXEON 系列，并在 AlInGaP 效率、高功率白光、暖白高功率 LED、手机闪光、日行灯、全 LED 头灯等多个节点上站在行业前沿。Audi A8 W12 的日行灯、Audi R8 的全 LED 头灯，也都可以看到它当年的技术影响力。

如果你经历过那个时代，就知道这些不只是“推出新产品”而已。

红光与黄光的高亮度化，让 LED 真正拿下了鼠标、信号显示、汽车尾灯与刹车灯市场。

高功率白光与封装突破，让 LED 不再只是 indicator，而开始有资格谈 illumination。

而高可靠性与车规导入，则让 LED 从电子零件变成汽车品牌设计语言的一部分。这些路径今天看起来仿佛是自然演进，但在当年，其实是少数几家公司用很长时间、很深技术，一步一步踩出来的。Lumileds 就是其中最关键的几家之一。

四、Philips 为什么能够华丽转身？Lumileds 是不能不提的底盘

后来 Philips 能从传统光源巨头成功转向 LED，很多人会先想到品牌、渠道、系统与全球市场能力。这些都对，但如果少了 Lumileds，这个故事并不完整。

Philips 在 2005 年收购 Agilent 在 Lumileds 的持股，本质上不是单纯多买一家元件公司，而是把一套足以支撑自己照明转型的底层光源技术平台收进体系。再到 2015 至 2016 年准备出售时，Philips 对 Lumileds 的定位，依旧是 LED 元件与汽车照明领域的重要核心资产。

从产业视角看，这是一个很高明、也很典型的动作。

当传统照明帝国意识到未来属于半导体照明，它若只守着灯泡、灯具、通路与品牌，迟早会被动；但若掌握了底层技术平台，它就有机会主导转型节奏。

所以我一直认为，Lumileds 对 Philips 的价值，从来不只是供应商，也不只是内部事业部。它曾是 Philips Lighting 华丽转身背后最重要的技术底盘之一。

五、真正的转折，不是技术衰退，而是角色改变

但一家公司的命运，往往不取决于它曾经多重要，而取决于母体还要不要它。

Philips 后来逐步聚焦健康科技，Lumileds 于是从曾经的转型引擎，慢慢变成可以拆分、估值、出售的资产。2015 年，GO Scale Capital 主导的财团原本拟以约 33 亿美元企业价值取得 Lumileds 多数股权；到了 2016 年改由 Apollo 接手时，整体企业价值已降到约 20 亿美元。

最值得玩味的，不只是价格差异，而是估值逻辑变了。

当一家公司是集团转型的技术引擎时，它带有战略溢价。

当它变成可出售的成熟资产时，它就很快会被改用财务资产的方式来衡量。这是 Lumileds 命运真正转折的起点。

六、为什么会被一再转手？因为 LED 产业 的价值中心已经变了

Lumileds 的故事，如果只写成“一家好公司命运多舛”，其实太浅。

更深一层是：它的命运变化，正是 LED 产业 价值中心变化的结果。

早期 LED 产业 的价值中心，在材料、外延、晶片、封装与可靠性。

中期开始，价值中心逐步向大规模制造、成本优化、供应链效率与全球扩产转移。

再往后，价值又向模块、系统、控制、品牌、场景与软硬件整合延伸。

Lumileds 在第一阶段极强，在第二阶段仍然重要，但到第三阶段，它虽然依旧有技术底蕴，却不再是唯一的舞台主角。它后来仍持续推出 CSP、矩阵车灯、SkyBlue、NightScape、microLED 等技术节点，说明它不是没有创新；问题在于，创新还在，不代表产业主导权还在。

也就是说，Lumileds 不是死于技术停滞。它更像是被整个产业的价值迁移，慢慢从舞台中央推到了侧翼。

七、从 PE 到地缘政治：Lumileds 为何两度卡在 CFIUS

Lumileds 最耐人寻味的地方，是它两次想卖给与中国资本相关的买方，都卡在美国国安审查。

第一次是 GO Scale 交易。2016 年 1 月，这笔交易终止，原因就是 CFIUS 疑虑未能排除。

第二次則是三安與 Inari 擬收購 Lumileds International。2025 年 8 月，雙方宣布交易；到了 2026 年 4 月，交易再次終止，公開披露指出，CFIUS 認定這項擬議聯合收購存在未解決的國家安全疑慮。

这件事非常关键。它说明今天的 Lumileds，已不能只用“一家 LED 公司”来看。

只要牵涉到美国技术、业务、客户、供应链，或具有战略含义的半导体光电能力，它就可能被放进更大的国家安全与科技竞争框架中审视。

这也是为什么我说，Lumileds 的命运不是单一企业命运，而是 LED 产业 从技术竞争，走向技术竞争加地缘政治竞争的缩影。

八、我为什么对这家公司特别有感？因为我看过它在不同时代的位置

今天回头写 Lumileds，我其实很难用纯学术口吻。因为我看过它在不同时代的样子。

我见过1998年槟城 HP Opto 工厂那种带着半导体工业荣光的气质。

我也在东贝担任采购时，见过它作为上游关键供应商的分量。

我更在“十城万盏”那种 LED 普及化与城市照明快速替代的历史现场，见过它如何把自身产品推进大规模应用。

这些经历让我更强烈地感受到：Lumileds 的可敬，不只是它曾经很强；而是它几乎参与了 LED 从技术萌芽、高亮度突破、汽车渗透、通用照明爆发，到大规模商品化的每一个关键阶段。

但也正因如此，它后来的命运更让人唏嘘。一个曾经帮助产业跨代、也帮助 Philips 转身的技术引擎，最后却没能稳稳掌握自己的归属。

九、以史为鉴：Lumileds 给今天所有企业的，不只是感慨，而是警示

我认为 Lumileds 最值得今天行业反思的，不是“英雄迟暮”，而是下面几件事。

第一，核心技术要持续升级，更要持续导入市场。

第二，元件龙头若无法逐步上移到系统、场景、标准或平台，最后很容易被重新估值为成熟制造资产。

第三，母公司若把关键技术引擎视为可处置资产，短期可能是正确财务决策，长期未必是最优战略选择。

第四，PE 可以重整资本结构，却不一定能重建产业主导权。

第五，在今天，半导体光电资产的跨境并购，已不能只用商业逻辑理解，而必须同时接受国安与地缘政治逻辑的检验。

结语

Lumileds 的故事，最让人感慨的地方在于：它不是一家没有技术的公司，不是一家没有历史的公司，也不是一家没有贡献的公司。

恰恰相反，它太有技术、太有历史、也太有贡献了。

它曾来自 HP 最强的光电血脉，曾替 Philips 撑起 LED 转型的底盘，曾定义红光、高亮度、功率封装与车用 LED 的时代，

也曾作为人才与管理母体，孕育并外溢出一整代影响欧美与中国 LED 产业的专业人士与管理层。

却也正因如此，在产业成熟、资本转向、主导权东移与地缘政治升温之后，它成为反复被交易、却始终难以安放的关键资产。

所以，Lumileds 留给今天所有企业的真正问题不是：为什么它这么坎坷？

而是：当一家企业曾因核心技术而伟大，它是否有能力让这项技术持续升级、持续主导市场，并且始终掌握在真正理解其长期价值的战略主体手中？

如果没有，那么今天的技术引擎，明天就可能只是下一轮交易里的待售资产。

从技术垄断、品牌高地到碎片化肉搏市场——这不只是 OSRAM / LEDVANCE 的故事，也是所有曾经成功企业都该读懂的一课

摘要

OSRAM 迎来品牌 120 周年。本文不仅回顾其从传统光源到数字光子学的技术演进，也从我曾参与 MLS 与 LEDVANCE 整合、并担任 LEDVANCE CEO 的视角，重新审视一个百年垄断者如何从卖方市场高地，走入碎片化竞争的全球肉搏战。这既是 LED 时代改写产业规则的必然，也是所有曾经成功企业都必须面对的反思：核心技术是否持续升级？Go-to-Market 是否尊重市场差异？品牌与渠道是否真正以市场为中心，而非始终以总部为中心。

导语

今年，OSRAM迎来品牌120周年。

对很多人而言，这是一家伟大的德国照明企业；

对照明行业而言，它几乎参与了整个现代光源工业的主线发展：从白炽灯、荧光灯、卤素灯，到 LED、汽车照明、红外与传感，再到今天所谓的数字光子学。OSRAM品牌最早注册于1906年，而公司则在1919年由 Auergesellschaft、Siemens & Halske 与 AEG 的灯业务合并而成。

但如果只把这篇写成一篇品牌庆生文，我觉得太可惜。

因为在我看来，OSRAM这 120 年真正值得回望的，不只是它曾经有多辉煌，而是它如何从一个技术领先、扩产就能增长的卖方市场寡占者，一步步走到后来必须在全球各地，与成千上万个竞争者短兵相接的局面。这其中当然有行业演进的必然，也有企业面对未来时的迟疑与失策。这不是旁观者的评论，而是我在参与 MLS 与 LEDVANCE的整合、并后来担任 LEDVANCE CEO 的过程中，亲眼看到的一段产业权力重排。

一、OSRAM的伟大，从来不只是“老牌子”，而是它曾经代表整个工业时代

OSRAM的起点，本身就带有很强的工业整合基因。品牌在1906年诞生，公司于1919年正式成立，背后是德国几家重要工业力量将灯泡业务整合到同一品牌之下。它不是单一创业故事，而更像是一个工业时代的集体作品。

也因此，OSRAM长期不是单纯卖产品，而是在卖一套非常完整的能力：材料、工艺、制造、品质、一致性、全球标准、品牌信用、渠道覆盖。

后来它一路将这种能力从传统光源延伸到更多技术节点。官方 120 周年页面梳理了其技术演进里程碑：从汽车灯、低压钠灯、荧光灯、氙灯、卤素灯，到红外 LED、彩色 LED、白光 LED，再到更后来的薄膜芯片技术与今天的 Digital Photonics。这意味着它曾经不只是照明企业，而是光科技演进的重要推动者。

所以，OSRAM值得尊敬的地方，不只是做大过，而是它曾经长期站在技术和市场的双高地上。

二、它与 Siemens 的关系，说明了它原本就站在工业体系的中心。

如果要理解 OSRAM 的历史，就绕不开 Siemens。

OSRAM成立时就与 Siemens有深度关系，并在很长一段时间内一直是西门子体系中的重要组成部分。直到2013年，Siemens推动 OSRAM 分拆上市，将80.5%的 OSRAM 业务分配给股东，自己保留17%股份，另有2.5%转入 Siemens Pension Trust。这是 OSRAM 从集团型工业体系走向独立资本市场的一个重要节点。

这个动作在当时未必代表衰退，反而更像是一种战略重估：当照明产业开始从传统光源进入 LED 与半导体时代时，OSRAM必须面对新的资本叙事、新的竞争逻辑，以及不同于 Siemens 工业母体的估值方式。分拆本身，是对未来的一次重新定义。

只是后来回头看，分拆只是开始，真正艰难的是：当一家公司离开过去那个稳定的大工业秩序后，它是否有能力适应一个更快、更碎片化、更残酷的新市场。

三、它与 GE、Siemens、AEG 的关系，提醒我们：照明曾经是一个高度寡占的工业体系

今天很多年轻从业者进入行业时，看到的是高度内卷、高度同质化、价格透明的 LED 市场，很难想象照明曾经是一个真正意义上的全球寡占产业。

OSRAM的成立，本身就是欧洲工业整合的结果；而更早期的照明产业，也与专利、技术、材料和跨国协议高度交织。这说明一件事：照明过去不是低门槛生意，而是一门由少数巨头主导、技术与产能深度结合的工业。

在那个年代，对龙头企业而言，增长很多时候不是“要不要重新定义自己”，而是“要不要扩产、如何扩产”。

因为市场本质上仍是供给稀缺的卖方市场。谁拥有技术、品牌、质量和全球渠道，谁就拥有更强的议价权。

这一点很重要。因为后来 OSRAM / LEDVANCE 所遭遇的尴尬，并不是普通企业都会经历的那种竞争，而是从寡占高地被拉入碎片化混战后的失重感。

四、真正改变一切的，不是某个对手，而是 LED 把整个行业规则改写了

我一直认为，OSRAM / LEDVANCE 后来所面对的局面，首先是行业演进的必然，其次才是企业自身的应对失误。

原因很简单：LED 改变的不只是光源本身，而是整个照明产业的底层规则。

当光源从传统工业制造转向半导体化、电子化、模块化以及全球供应链协同之后，产业门槛的构成就发生了变化。市场不再只奖励那些拥有深厚历史资产、标准能力与品牌信用的大公司，也开始奖励更快的供应链、更低的成本结构、更高频的产品迭代，以及更贴身的本地竞争能力。

这意味着，原本属于少数巨头的游戏，逐渐变成大量企业都能进入的战场。

不是 OSRAM 突然变差了，而是它原本那套赢法，不再自动成立了。

五、但把一切都归因于“时代变了”，也并不公平

如果只是行业宿命，那么所有传统巨头都应该以同样方式失速。

但事实并非如此。

所以我认为，OSRAM / LEDVANCE 的经验真正值得所有企业反思的，不只是市场从卖方转成买方，不只是竞争者从几十家变成几千家，而是：曾经的寡占者，是否仍然有能力持续升级核心技术、持续储备下一代技术，并且有决心把这些能力导入市场；同时，它的 Go-to-Market 模式，是否真的愿意尊重各个市场的差异，而不是始终以总部视角看待全球。

这才是关键。

很多企业不是看不见未来，而是以为自己过去成功的方式，仍然足以管理未来。

六、第一个教训：核心技术不能只“领先过”，而必须能够持续升级、持续商业化。

OSRAM的技术实力毋庸置疑。从官方120周年资料可以看出，它在多个时代都站在重要技术节点上：汽车光源、放电灯、卤素灯、红外 LED、白光 LED、薄膜芯片、数字光子学。它从来不是没有技术的公司。

但对任何寡占者来说，问题往往不是“有没有领先过”，而是：

• 能不能持续推进下一代技术；
• 能不能把技术储备转化为真正的市场方案；
• 能不能在新技术刚成形时，就建立相应的组织能力、供应链能力与商业化能力。

很多企业到后来会出现一种典型情况：有研发、有专利、有实验室、有工程能力，却没有足够果断地把它推进市场，或没有及时把组织调整到能够承接新技术的节奏。

这种问题在顺风时期并不明显，一旦市场规则发生变化，就会变得非常致命。

因为技术领先如果不能转化为新的商业秩序，就只能停留在荣誉簿上。

七、第二个教训：真正拖垮寡占者的，往往不是竞争者变多，而是决策模型失效

这是我认为最值得今天所有大型企业反思的一点。

很多全球性龙头在强势时期，都建立了一套“总部最懂”的体系：

• 总部定义产品
• 总部定义品牌
• 总部定义价格带
• 总部定义节奏
• 地方市场主要负责执行

这套模型在供给稀缺、技术节奏相对稳定、品牌权威很强的年代，是有效率的。

但一旦市场进入高度分化的竞争阶段，它就会逐渐失灵。

因为不同市场的现实根本不一样：

• 对价格的敏感度不一样
• 对品牌的理解不一样
• 通路的权力结构不一样
• 对交期、服务、SKU、促销与定制化的要求不一样
• 竞争者密度与打法也完全不同

如果仍然以总部为核心，而不是以市场为核心，企业就很容易低估本地需求、拖慢响应速度，最终问题往往不是某一个决策错了，而是整个决策模型开始失效。

很多寡占者真正的问题，不是没有资源，而是反应速度已经慢于市场变化的速度。

八、第三个教训：品牌与渠道，不能只追求全球一致，更要追求市场有效

这一点在照明这种跨区域、跨渠道、跨应用场景的产业里尤其明显。

品牌不是总部自己认为清楚就够了，品牌真正的价值在于市场是否愿意为它付费、是否愿意让它进入采购决策，以及是否愿意在渠道端为它保留位置。

OSRAM在历史上很成功地经营了全球品牌，也通过不同市场路径来扩张。比如在北美，通过 SYLVANIA 建立强势存在；LEDVANCE 今日也仍把 SYLVANIA 品牌故事作为北美的重要资产之一。

再看日本市场，三菱电机在1980年代末曾与 OSRAM成立两家合资公司，一家偏生产，一家偏销售。到2012年三菱电机重整照明业务时，官方也明确说明，这些合资的初衷就是结合 OSRAM 的技术优势与三菱电机的销售优势；即使在重组后解除合资关系，双方仍保留销售与生产合作。这其实就是一种很典型的“尊重市场结构”的做法。

这些案例都说明：真正强大的全球品牌，不是把一套总部想象复制到全世界；

而是清楚知道，全球可以有统一方向，但市场必须有本地打法。

品牌若只对总部负责，渠道若不对市场现实负责，再好的历史资产，最后也可能被一点一滴耗掉。

九、第四个教训：总部控制力，不等于市场掌控力

很多企业在巅峰时期，很容易把两件事混为一谈：我能控制全球组织，等于我能掌控全球市场。

但这两件事其实差很多。

总部控制力强，可能代表流程严谨、品牌一致、风险可控；但市场掌控力强，则意味着前线足够敏捷，能够快速判断需求、调整产品组合、重设价格带、改变渠道策略，甚至重写打法。

在市场单一、节奏较慢的时代，两者可能重合；但在今天这种碎片化、区域化、即时竞争的时代，两者往往是冲突的。

如果总部仍然习惯把所有决策向上集中，地方市场就会越来越像执行单元，而不是作战单元。

一旦前线不敢决策、没有授权、不被尊重，再强的品牌也很难打赢贴身肉搏战。

所以，对所有曾经的寡占者而言，真正需要建立的不是更强的总部控制，而是更高质量的全球—本地协同能力。

十、LEDVANCE的分拆与出售，不只是交易，更是产业权力重排

2016年，OSRAM监事会批准出售一般照明灯与光源业务，也就是后来大家熟知的 LEDVANCE，买方为由 IDG、MLS 与义乌国资组成的中国财团，交易价格超过4亿欧元。OSRAM当时对外表述得很清楚：这是其向高科技公司转型的重要里程碑。

后来到了2018年，LEDVANCE公告 MLS 成为唯一股东，并提到 LEDVANCE 源自 OSRAM 的一般照明业务，业务遍及140多个国家与地区。

如果只从财务交易来看，这是一笔分拆出售；但如果从产业史来看，它远不止如此。

它其实代表了一次重要的全球权力转移：一边是欧洲老牌工业体系，把一般照明这个竞争越来越激烈、毛利越来越薄的业务剥离出去；

另一边则是中国企业凭借制造、供应链、成本效率与资本能力，接住全球品牌、渠道和市场体系，开始重组一般照明的竞争格局。

我后来身在其中，更强烈感受到：这不是某一家公司突然不行了，而是原本属于一个时代的赢法，正在被另一套能力体系取代。

十一、所以，这到底是行业必然，还是自身失策？

我的答案是：首先是行业的必然，其次是企业的失策。

必然在于，LED 确实把照明产业从高门槛、相对寡占的工业，推向更电子化、更供应链化、更全球化、也更中国化的竞争结构。这个大方向很难逆转。

但失策也确实存在。

失策不一定是做错某一个单一决策，而是：

• 没有足够快地重新定义技术的商业化方式；
• 没有足够早地接受市场权力已经从总部转向前线；
• 没有足够深入地重建 Go-to-Market 模型；
• 没有在品牌与渠道上真正做到以市场为中心；
• 仍然试图用主导旧时代的方法，去管理一个已经完全不同的新时代。

这才是最大的问题。

十二、这不只是 OSRAM 的故事，而是所有曾经成功企业的共同课题

我想，这篇文章写到这里，重点已经不只是回顾 OSRAM 120 年。

更重要的是，它给所有企业——尤其是那些曾经在某个领域拥有寡占地位、技术高地或品牌高地的企业——敲响了一个警钟：寡占者最大的风险，不是失去昨天的优势，而是把昨天的优势误认为是明天的能力。

你曾经的成功，可能来自技术；但未来的成功，可能更取决于技术升级、组织升级、市场升级是否同步完成。

你曾经靠总部主导赢得效率；但未来如果仍然低估市场差异、低估前线判断、低估本地需求，你的决策模型本身就可能成为竞争劣势。

你曾经靠品牌与渠道建立护城河；但如果品牌不能重新定义、渠道不能贴近市场，护城河也会被时间慢慢填平。

结语

OSRAM120周年，值得致敬。

因为它不只是欧洲工业的一段历史，也是一部完整的照明产业进化史。从1906年品牌注册，到1919年公司成立，再到后来与 Siemens 的长期关联、2013年分拆上市、2016年出售 LEDVANCE，以及在不同市场通过 SYLVANIA、与三菱电机合作等方式走向全球，这家公司的一生几乎浓缩了现代照明产业的主要脉络。

但我更想说的是：真正值得今天的企业学习的，不是 OSRAM 曾经有多强，而是它的曲折提醒了我们——

没有任何一家企业，可以永远用主导旧时代的方法，继续赢得新时代。

这，才是 OSRAM 120年留给行业最重要的启示。

From HP Opto and Philips to private equity and geopolitics:

摘要

Lumileds is more than a company with a complicated ownership history. It is a lens through which to view the shifting center of power in the LED industry. Born from the HP Opto lineage, it helped define major eras in red LEDs, high-brightness performance, power packaging, and automotive applications, while also supporting Philips’ strategic transition into LED lighting. But as the industry matured, value migrated from core device technology toward scale, cost, systems, and ecosystems. Capital restructuring and geopolitics then further reshaped the company’s fate, turning Lumileds from a strategic engine into an asset repeatedly put up for sale. Its story raises a larger question for every technology-driven company: can core technical capability remain strategically owned, continuously upgraded, and continuously translated into market control — or will it eventually become just another asset on the transaction table?

A company that once helped define an era, only to become an asset for sale**

In 1998, I visited the HP Opto factory in Penang.

Even now, I still remember the feeling.

That was not the LED industry people know today — an industry reshaped by price competition, manufacturing scale, and supply-chain efficiency. Back then, LED still carried a very strong semiconductor-industrial character: clean, restrained, precise. Behind it stood real hard power — materials, process engineering, packaging, reliability — and also the quiet confidence of a technology era that still believed deep engineering could shape the future.

Later, when I was responsible for procurement at Everlight, Lumileds was one of our most important suppliers in Taiwan. Later still, in China’s “Ten Cities, Ten Thousand Street Lamps” (十城萬盞）street-lighting programs, I again encountered Lumileds products in real projects and real deployment decisions.

So in a sense, I did not come to know Lumileds only through headlines or corporate history. I encountered it repeatedly, in different historical positions, at different stages of the industry.

That is why, when I look back at Lumileds today, I find it difficult to see it as merely the story of one company’s ups and downs.

To me, it is more like a coordinate point in the industry’s collective memory.

If you follow its journey carefully, you can almost retrace the most important shifts in the LED business over the past three decades: from American optoelectronics labs to European strategic integration, from the golden age of high-brightness LEDs, power packaging, and automotive applications to the rise of Asian manufacturing, and then on to capital restructuring, private equity ownership, and the new reality of geopolitics and national security review.

Some companies grow inside an era. Lumileds, in my view, is one of the rare companies that lived through the turning of an entire era.

And what it leaves behind is not just a technical legacy. It leaves a question that every technology-driven company must eventually face:

How does a technology engine become a saleable asset?

1. Lumileds did not begin as a lighting company. It began in semiconductor optoelectronics.

Many younger people in the industry instinctively think of Lumileds as simply a Philips LED company. That is not wrong. But it is not enough.

The roots of Lumileds are not really in traditional lighting. They are in HP’s optoelectronics and semiconductor heritage.

This matters. Because it means Lumileds did not grow out of the conventional logic of lamps, fixtures, channels, and lighting engineering. It grew out of a different logic altogether: materials, epitaxy, chips, packaging, reliability, and productization.

Many lighting companies first understood “light” and then learned “LED.” The HP–Agilent–Lumileds line followed the reverse path: it first understood semiconductor light, and only then expanded into lighting, automotive, and broader real-world applications.

That difference in origin shaped everything. It shaped the kind of problems the company solved. It shaped the kind of talent it built. And it shaped the role it played in the early LED industry.

Lumileds was never just another supplier. At its core, it was a platform company born from deep technology. And that is exactly why its later fate became so symbolic.

2. HP Opto was not just a predecessor. It was one of the early talent nurseries of the LED industry.

If we describe HP Opto merely as the predecessor of Lumileds, we still underestimate its historical importance. To me, HP Opto was more like a practical academy for the early LED industry.

It trained not only engineers, but an entire generation of people who understood how to move from material science to manufacturing, from chip to package, from reliability to product, and from product to large-scale market adoption.

That kind of capability was rare. LED was never the kind of technology that could succeed simply because a laboratory proved it could work. It had to cross many thresholds: scientific feasibility, manufacturability, reliability, consistency, customer trust, application fit, and economic viability.

What made the HP Opto lineage special was not only that it produced technology. It produced people who knew how to industrialize technology.

Over time, talent from that lineage did not remain within one company. It spread outward — into Lumileds, Cree, Bridgelux, and many other LED and lighting businesses across Europe, the United States, and China.

That is why I do not see HP Opto merely as a historical ancestor. I see it as one of the early talent and management mother-bodies of the LED era.

If Lumileds later became an important technology engine for Philips’ transformation, then HP Opto was, in many ways, one of the industry’s earliest academies of execution.

It did not only produce products. It produced methods, discipline, industrial DNA, and a generation of professionals who would shape the next chapters of the LED business.

3. Lumileds did not merely lead. It helped define an era.

Younger professionals entering the industry today may find it difficult to imagine how strong Lumileds once was. It was not just another company making LEDs. For a significant period of time, it was one of the companies defining the path of the industry itself.

Its technologies influenced several key turning points: High-brightness red and amber LEDs helped LEDs take over applications such as mice, signal indication, automotive tail lamps, and brake lights. Power white LEDs and packaging breakthroughs helped LEDs move beyond indication and earn the right to be taken seriously for illumination. Automotive-grade reliability and integration helped turn LEDs from mere electronic components into part of automotive brand language and design identity.

If you lived through those years, you know these were not just product launches. They were moments when a handful of companies, through deep technical investment and long-term execution, carved out entirely new application paths.

Today, it seems obvious that LEDs belong in automotive lighting, in general illumination, in street lighting, in architectural lighting. But none of that was inevitable. Those roads were built. And Lumileds was one of the companies helping to build them.

4. Why was Philips able to turn so successfully toward LED? Lumileds was one of the crucial foundations.

When people think about Philips’ transition from a traditional lighting giant into the LED era, they often think first of brand strength, global channels, systems capability, and market reach.

All of that matters. But the story is incomplete without Lumileds. What Philips understood, earlier and more clearly than many others, was that if the future of lighting belonged to semiconductor light sources, then controlling fixtures, channels, and branding would not be enough. It also needed access to the underlying source technology platform.

This is what made Lumileds strategically important. Lumileds was not just a component supplier sitting somewhere inside the broader Philips structure. It was one of the technical foundations that allowed Philips Lighting to move from legacy light sources into the LED age with real credibility and momentum.

In other words, Philips’ LED transition was not only a commercial repositioning. It was also a reassembly of technical capability. And Lumileds was one of the key pieces in that reassembly.

5. The real turning point was not technical decline. It was a change in role.

A company’s fate is often determined not by how important it once was, but by whether it is still seen as part of the future. That, in my view, was the decisive turning point for Lumileds.

The problem was not that it forgot how to innovate. The problem was that its role changed.

When it was seen as a strategic engine, it carried strategic premium. When it became a separable asset, it began to be judged through a different lens: valuation, transaction structure, buyer suitability, and exit logic.

This is where many technology companies begin to lose control of their destiny. Not because the technology suddenly becomes weak. But because the company ceases to be treated as a strategic capability and starts being treated as a financial object.

From strategic asset to financial asset. From engine of transformation to asset available for disposal. That shift is subtle at first. But once it happens, it changes everything.

6. Why was Lumileds sold and resold? Because the center of value in the LED industry moved.

If we reduce the Lumileds story to “a great company with bad luck,” we miss the deeper point. Its fate changed because the center of value in the LED industry changed.

In the early phase of LED, value was concentrated in materials, epitaxy, chips, packaging, and reliability. Whoever mastered those layers had technical authority. Whoever had technical authority could occupy a commanding position in the value chain.

But as the industry matured, value began to migrate. It shifted from core technical breakthroughs to manufacturing scale. From manufacturing scale to cost structure and supply-chain efficiency. And then further still, toward modules, systems, controls, branding, scenarios, and integrated hardware-software ecosystems.

That shift is profound. It means that a company that was extraordinarily powerful in the “component age” may not remain equally central in the “systems age” or the “platform age.”

Lumileds did not stop innovating. That is not the point. The point is simpler, and harsher: Innovation can continue, while industry control quietly moves elsewhere.

This is one of the hardest truths for old technology leaders to accept. Sometimes you do not become weak. Sometimes the center of gravity simply moves.

7. From private equity to geopolitics: why Lumileds was blocked twice

One of the most revealing aspects of the Lumileds story is that attempts to sell it to buyers tied to Chinese capital were blocked twice by U.S. national security review. At that point, Lumileds was no longer just an LED company. It had become a sensitive optoelectronic asset.

Once a company touches U.S. technology, U.S. operations, U.S. customers, or strategically relevant semiconductor capabilities, the question is no longer just: who wants to buy it, and at what price?

The question becomes: how will it be interpreted inside a broader framework of national security, technology competition, and supply-chain control?

That is why Lumileds is so important as a case study. Its story cannot be understood only at the level of corporate management. It belongs to a larger narrative: from technology competition, to technology plus capital restructuring, to technology plus capital plus geopolitics.

And that same logic is now visible far beyond lighting.

8. Why does this company feel personal to me? Because I saw it in different positions across different eras.

When I write about Lumileds today, it is difficult for me to write as a detached observer. Because I saw it in different forms, at different moments.

I saw the Penang HP Opto factory in 1998, when LED still carried that distinctive semiconductor-industrial seriousness. I saw Lumileds again as a major supplier when I was in procurement at Everlight. And I saw it again in the phase when LED began moving rapidly into larger-scale urban lighting and infrastructure programs.

So for me, Lumileds is not just a company name in an article. I have seen it as a technology source. I have seen it as a supplier of consequence. I have seen it participate in the transition from high-end technical capability to large-scale real-world adoption.

That is why its story carries weight. It was not a latecomer riding a trend. It was there across nearly every important phase of the LED era: early technical formation, high-brightness breakthroughs, automotive penetration, general-lighting expansion, and finally mass-market commoditization.

A company like that ending up repeatedly traded, repeatedly reviewed, and never quite settled in ownership — that is bound to provoke reflection.

9. What should the industry learn from Lumileds?

If we revisit Lumileds today, the real takeaway is not merely that its story is unfortunate. It is that it offers several warnings to every technology-driven business.

First, core technology must continue to advance — but it must also continue to enter the market effectively. Technical leadership that does not translate into next-generation market relevance will eventually thin out.

Second, component leaders that fail to move upward toward systems, platforms, scenarios, standards, or ecosystem control are likely to be revalued as mature manufacturing assets.

Third, when a parent company begins to see a technical engine as a disposable asset, the short-term financial logic may be sound, but the long-term strategic cost may be much higher than it first appears.

Fourth, private equity can repair balance sheets, restructure portfolios, and improve financial flexibility — but that is not the same as restoring a company’s historical place in the industry.

Fifth, critical semiconductor and optoelectronic assets can no longer be understood through commercial logic alone. They are increasingly shaped by three forces at once: technology logic, capital logic, and national-security logic.

Ignore any one of them, and you may fundamentally misread where you stand.

Conclusion

What makes the Lumileds story so striking is this: It was not a company without technology. It was not a company without history. It was not a company without contribution.

Quite the opposite. It carried some of the strongest optoelectronic DNA of the HP era. It helped support Philips’ LED transformation. It helped define the eras of red LEDs, high-brightness performance, power packaging, and automotive LED adoption. And it also served as a talent and management mother-body whose influence spread far beyond one corporate boundary.

And yet, precisely because of all that, it eventually became a key asset that was repeatedly sold, repeatedly re-evaluated, and repeatedly difficult to place. So the real question Lumileds leaves behind is not: Why was its journey so turbulent?

The real question is: When a company becomes great because of core technology, can it continue upgrading that technology, continue shaping the market with it, and keep that capability in the hands of a strategic owner who truly understands its long-term value?

If not, then today’s technology engine may become tomorrow’s asset for sale.

From technological leadership and brand power to fragmented, close-range competition — this is not only the story of OSRAM/LEDVANCE, but a lesson every former market leader should study.

Introduction

This year, OSRAM marks its 120th anniversary.

For many, OSRAM is a great German lighting company. But in my view, seeing it only as a “legacy lighting brand” does not go nearly far enough.

OSRAM was never just a company, and never just a brand. It participated in almost the entire arc of modern lighting history: from incandescent lamps, fluorescent lamps, and halogen, to LED, automotive lighting, infrared, sensing, and today’s broader world of photonics and digital light technologies.

Yet what is even more worth revisiting than its history is the path it has taken over the past decades: from being a technology-led oligopolist in a seller’s market — where growth could often be unlocked simply by expanding capacity — to becoming a company forced to fight, market by market, against thousands of competitors in highly fragmented global battlegrounds.

Was that outcome inevitable because the industry changed? Or was it also the result of strategic misjudgment?I believe the answer is both.

But the more important question for today’s business leaders is this: when a company has spent decades at the top, can it still reinvent the way it wins?

I was not an original OSRAM employee. But I was deeply involved in the LEDVANCE chapter — from my role as Managing Director within MLS to later serving as CEO of LEDVANCE. I witnessed, from inside the process, how industrial power, brand architecture, channel logic, and competitive rules were all being rewritten.

So this is not intended to be a simple anniversary tribute. Rather, I want to use OSRAM at 120 as a lens to reflect on the evolution, the challenges, and the lessons of a century-old oligopolist.

1. OSRAM’s greatness was never just that it was “old” — it once represented an entire industrial era

The name OSRAM carries the imprint of the industrial age. It was not a startup story in the modern sense.

It was born out of a period of deep European industrial consolidation. And because of that, OSRAM never sold only “lamps.” It sold a complete industrial capability:

• materials expertise
• manufacturing scale
• process know-how
• quality control
• standards and certification
• brand trust
• global distribution reach

In the era of traditional lighting, these capabilities created a very high barrier to entry. Not everyone could do it. Fewer still could do it reliably, globally, and at scale. That is why OSRAM was not simply a “large company.” It occupied a genuine industry high ground.

For a long time, it stood for: stronger technology, more stable quality, broader product portfolios, and greater authority in the market. This is why, when we look at OSRAM today, we should not start with what it later struggled with.

We should first acknowledge what it once truly was: one of the defining winners of its age.

2. From seller’s market to buyer’s market: many giants did not suddenly become weak — the rules of the game changed

To understand the later challenges faced by OSRAM and LEDVANCE, we must begin with one simple truth: this was, first of all, an industry transition that was structurally inevitable.

In the era of traditional light sources, the industry had several defining characteristics: First, technological and manufacturing barriers were high.

Second, brands and channels were highly concentrated.

Third, demand grew steadily over long periods.

Fourth, the number of global competitors was limited, and the market remained, in essence, a seller’s market.

In that context, if a market leader had strong technology, a trusted brand, and broad channels, growth often really was a matter of capacity expansion.

Then LED changed everything. LED looked like a source substitution, but in reality it rewrote the foundations of the entire industry:

• technology diffused faster
• manufacturing became more replicable
• products became more commoditized
• costs declined more quickly
• supply chains became more global — and more China-centered
• the number of market participants surged
• channels became more fragmented
• price competition became more direct and more frequent

What had once been a long-distance race among a handful of dominant players gradually turned into a crowded battlefield with thousands of companies competing across markets.

So what OSRAM and LEDVANCE later faced was not simply “more competition.” It was something deeper: the lighting industry collapsing from a technology-driven seller’s market oligopoly into an overcompetitive buyer’s market.

That was not caused by any one company alone, nor by the arrival of any single competitor. It was the result of a structural rewrite of the industry itself.

3. But it would be too easy — and too unfair — to explain everything by saying “the times changed”

If this were only fate, then all legacy giants should have lost momentum in exactly the same way. They did not.

Which brings us to the second layer of analysis: industry change may have been inevitable, but how a company responds determines whether it adapts or gets trapped.

In my view, the most valuable lesson from OSRAM and LEDVANCE is not merely that margins were squeezed or competitors multiplied. It is that every former oligopolist must ask itself a harder set of questions:

• Has core technology continued to evolve?
• Has the next generation of technology been actively built and prepared?
• Were those technologies decisively brought to market?
• Has the organization been upgraded to match the new pace of competition?
• Has the Go-to-Market model truly respected the differences between markets?
• Have brand-building and channel-building been centered on markets — or have they remained centered on headquarters?

These are the questions that matter most. Many companies do not fail because they cannot see the future. They fail because they assume that the logic that made them successful in the past will remain sufficient in the future.

4. First lesson: technology cannot merely have been leading once — it must keep evolving and keep converting into market power

One of the easiest traps for an oligopolist is to treat technological leadership as a static asset. Being ahead once does not mean remaining ahead. Having the capability to develop technology does not mean having the speed to commercialize it. Owning R&D capacity does not mean having the organizational ability to turn it into new market order.

The real issue is never simply whether a company has technology. The real issue is this: Can it continuously push the next generation forward, and can it decisively translate technical reserves into market capability?

Many former leaders did not lack R&D, patents, labs, or engineering talent. What they often lacked was:

• the willingness to accelerate the next wave
• the speed to bring emerging technologies into the market
• the organizational structure to support the transition
• the ability to form a new business system around the new technology

This is why some companies later seem, from the outside, to be neither weak nor obsolete — only late. Not incapable, but too slow. Not technically empty, but commercially under-converted.

For any former oligopolist, technology leadership that does not keep renewing itself — and does not keep converting into market advantage — eventually turns from moat into museum piece.

5. Second lesson: what often drags down former oligopolists is not the number of competitors, but the collapse of their decision model

This, to me, is one of the most important lessons. Many global leaders build a highly centralized operating logic during their strongest years:

• headquarters defines the products
• headquarters defines the brand
• headquarters defines the pricing architecture
• headquarters defines the pace
• regional teams execute

In an era of supply scarcity, strong brands, and relatively stable product structures, this model can be highly efficient. But once markets start to differentiate sharply, it begins to fail.

Because market reality is never uniform:

• price sensitivity differs
• buying behavior differs
• channel power structures differ
• engineering vs. retail mix differs
• competitor density differs
• service expectations differ
• SKU complexity differs
• decision chains differ

If a company continues to govern everything from a headquarters-centered logic, three things tend to happen: First, local demand gets underestimated. Second, decision speed falls behind market speed. Third, regional teams gradually lose their fighting capability.

At that point, the issue is no longer one bad product or one weak market. The issue is that the entire decision model has become unfit for the environment.

Many former oligopolists do not lose because they lack resources. They lose because: their response speed becomes slower than the speed of market change.

6. Third lesson: brand and channel strategy cannot only answer to headquarters — they must answer to market outcomes

This is another area where multinational companies often misread the challenge. Headquarters tends to look at branding through the lens of consistency. Markets judge brands through the lens of effectiveness. Headquarters wants one unified narrative, one unified image, one unified asset system.

But markets ask different questions:

• What does this brand actually mean here?
• Does it influence specification or procurement?
• Does it help channels make money?
• Is its positioning clear enough across price segments?
• Can it still compete meaningfully against local rivals?

The same brand can mean very different things in different markets. In mature markets, it may signal reliability and quality. In emerging markets, it may first be compared on price. In project markets, it may need to win through technical support and delivery performance. In retail, it may need to fight through shelf presence, promotions, e-commerce traffic, and consumer education.

That is why brand strategy cannot only pursue global uniformity; it must also pursue local effectiveness.

The same applies to channels. Strong channel-building is not about replicating the headquarters’ ideal blueprint around the world. It is about respecting the actual power structure, economics, and operating logic of each market.

Global brands absolutely need unified direction. But market competition must respect local reality.

If the brand answers only to headquarters, and the channel does not answer to the market, then even the strongest historical asset will slowly be consumed.

7. Fourth lesson: stronger headquarters control does not necessarily mean stronger market control

At their peak, many companies naturally fall into a dangerous assumption: the more tightly we control the organization, the more securely we control the market.

But these are not the same thing. Headquarters control may produce process discipline, brand coherence, and risk management. Market control, however, is reflected in something else:

• whether the front line is respected
• whether local teams are empowered
• whether products can be adjusted quickly
• whether price and channel strategies can react in real time
• whether branding can adapt to local competition
• whether decisions are made where the competition actually happens

When markets become more fragmented and more immediate, a company that keeps pulling decisions upward turns local teams into execution arms instead of fighting units.

Once the front line loses authority, flexibility, and speed, a familiar pattern appears: internally, the company still looks orderly; externally, it becomes increasingly hard to win.

For every former oligopolist, the goal should not be ever-stronger central control. It should be a higher-quality model of global-local coordination. Headquarters should own direction, platforms, principles, and capital allocation.

The market front line must own sufficient definition power, reaction speed, and battle-readiness.

8. The LEDVANCE carve-out and sale were not just a transaction — they marked a reordering of industrial power

The capital moves that followed made these structural shifts even more visible. The carve-out, the sale, and the eventual transfer of much of the general lighting business into a new ownership and manufacturing logic were not isolated financial events. They reflected a broader reorganization of value in the global lighting industry.

In one sense, this was the handoff between two capability systems: On one side stood the accumulated strengths of the traditional European industrial model — brand, product definition, customer relationships, global organizational experience, channel architecture. On the other stood the capabilities that Chinese companies built during the LED era — manufacturing efficiency, supply chain control, cost competitiveness, speed, and capital efficiency.

So this should not be understood merely as “selling a business.” It was, more fundamentally: a formal passing of the baton between the old order and the new order in lighting.

Having later operated within that reality, I felt this very clearly: This was not simply a case of one company becoming weak. It was a case of an old winning logic no longer being sufficient in a new competitive age.

9. So was it industry inevitability, or strategic misjudgment?

My answer remains: first, it was industry inevitability; second, it was corporate misjudgment.

The inevitability lay in the fact that LED pushed lighting out of an era defined by technological oligopoly, concentrated brands, and relatively stable channels, into one defined by electronics, supply chains, globalization, fragmentation, and speed.

But the misjudgment was also real. Not necessarily as one single wrong decision, but as a pattern:

• not upgrading the technology path fast enough
• not recognizing early enough that market power was shifting toward the front line
• not truly rebuilding the Go-to-Market model
• not making brand and channel strategy genuinely market-centered
• continuing to manage a new market with methods designed for an old one

That, in my view, is the deeper failure mode.

10. This is not only OSRAM’s story — it is the shared challenge of every former winner

At this point, the real subject is no longer only OSRAM at 120. The deeper point is what this story says to every company that once held a technology high ground, a brand high ground, or a channel high ground: the greatest risk for an oligopolist is not losing yesterday’s advantage; it is mistaking yesterday’s advantage for tomorrow’s capability.

A company may once have won through technical superiority. Tomorrow it may need organizational superiority, market adaptability, and system capability as well.

It may once have won through central coordination and scale efficiency. Tomorrow it may lose because it ignored local market intelligence and local competitive reality.

It may once have built a moat through brand and channels. Tomorrow it will have to prove again that the brand still matters, and that the channels still work in a transformed market structure.

For every business, the underlying question is the same: Are you willing to accept that a new market requires a new way of winning?

Conclusion

OSRAM at 120 deserves respect. Not only because it is a historic company, but because its journey is a condensed history of the modern lighting industry itself.

What deserves the most attention today, however, is not only how strong OSRAM once was. It is what its path now teaches us: no company can keep winning a new era with the methods that defined the old one.

That, to me, is the most important lesson OSRAM’s 120 years leave to the industry.

Short Summary

Written on the occasion of OSRAM’s 120th anniversary, this article goes beyond celebrating the company’s history. Drawing from my own experience across the LEDVANCE chapter — from Managing Director within MLS to CEO of LEDVANCE — it reflects on how a century-old market leader moved from technological and brand high ground into fragmented global competition. LED reshaped the rules of the lighting industry; that part was structural and inevitable. But what deserves deeper reflection is how technology renewal, organizational renewal, and market renewal often fail to happen in sync — especially when the Go-to-Market model remains headquarters-centered long after markets have become deeply local.

不是最亮的空间 往往是最舒服的空间

有些空间，第一眼并不震撼。

它没有特别夸张的灯光语言，没有刻意强化的设计存在感，也不是那种一拍照就很容易出片的地方。

但奇怪的是，当你真的坐下来、待上一阵子，反而会慢慢感受到一种难得的舒服：

不累。
不躁。
不想赶快离开。
甚至，会想再多待一会儿。

---

我们大概都去过这样的地方。

可能是一家餐厅、一间咖啡馆、一个酒店大堂，也可能只是某个不算起眼的会客空间。它未必豪华，未必惊艳，但人一进去，整个身体和注意力会慢慢放松下来。

你不一定立刻说得出原因，却能很明显地感受到：这里很顺，很稳，很适合停留。

---

很多时候，这种“待得住”的感觉，未必来自装修多高级、材料多昂贵，
而是因为那里的光，

没有一直打扰你。

---

一、我们常常被“第一眼效果”误导了

今天很多空间，尤其在社交媒体、样板间、展示空间的影响下，越来越追求第一眼的视觉刺激。

要有亮点。
要有反差。
要有记忆点。
最好一进门就让人“哇”一下。

这种追求并没有错。空间需要辨识度，商业场所也需要吸引力，设计本来就有表达的任务。

问题在于， 第一眼有效，不等于长时间舒服。

有些空间很适合拍照，却不适合久坐。
有些空间一开始很有气势，但待久了之后，眼睛会累，注意力会散，人也会莫名烦躁。
还有些空间，明明每个局部都很“用力”，但整体却让人无法真正放松。

这其实是今天很多空间共同的问题：太重视瞬间刺激，却忽略了停留体验。

但对大多数真实场景来说，人不是只看一眼就走。
他可能要坐下来吃饭、交谈、工作、休息、等候、阅读，甚至只是发一会儿呆。

这时候，光的价值就不再只是“有没有存在感”，
而是“会不会一直消耗人”。

---

二、真正决定停留感的，不只是亮不亮

很多人谈到舒服的空间，第一反应是亮度，或者色温。

但真正影响停留感的，往往没有那么简单。 一个空间让人愿不愿意待下来，常常取决于几件更细微、却更根本的事。

---

首先， 是光是否稳定。.

这里说的稳定，不只是技术层面的稳定，而是一种整体感受上的稳定。空间里的光不要忽明忽暗，不要东一块强、西一块弱，不要每个角落都在抢注意力。

当一个人的视线和神经系统不需要反复适应，他才比较容易安定下来。

---

其次， 是视线落点是否舒服。.

人进入空间，不是只看桌面，也不是只看某一盏灯。我们会看人、看墙、看前方、看远处，也会不自觉地在不同层次之间移动视线。

如果光的分布让视线总是在不舒服的亮点之间跳动，或者总要绕开刺眼的地方，人就很难真正放松。

---

再來， 是空间里有没有过多刺激与分心。.

不是越多亮点越高级，也不是越多层次越舒服。

有时候，太多强调、太多装饰性的发光元素、太多急着表现自己的照明手法，反而会把空间弄得很吵。

这种“吵”，不一定是声音，而是一种视觉上的持续打扰。

---

最后， 是人在里面是否需要一直“用力看”。.

很多人没有意识到，眼睛累，不一定是因为太暗；

很多时候，是因为空间让你必须一直调整自己。 调整焦点、调整视线、调整明暗适应、调整注意力分配。

这些看似细小的消耗，一旦持续发生，就会直接影响人愿不愿意久待。

---

所以，真正舒服的光，未必最亮，也未必最有戏，而是它让人不需要一直和空间对抗。

---

三、好光的价值，是让人不知不觉放松下来

我一直觉得，真正好的照明，有一种很重要但常被忽略的能力：它不一定让你立刻赞叹，却会让你慢慢松下来。

这种松，不是昏暗，不是无聊，也不是没有设计。

而是它没有用过度刺激去占据你的感官，于是你的注意力可以回到活动本身，回到眼前的人，回到真正重要的事。

---

当光做对了，人会少一些疲劳。你不需要一直眯眼，不需要反复调整视线，也不会总觉得哪里不对劲却又说不上来。

这种低负担，对餐饮、酒店、办公、零售、居住，全部都很重要。

---

当光做对了，人也会少一些压迫。

有些空间其实不差，但就是有一种说不出的紧。 可能是亮点过猛，可能是反差过重，可能是背景太黑、前景太亮，也可能是整个空间的视觉节奏太急。

久了之后，人很容易心烦、分心，甚至想赶快离开。

---

当光做对了，还会少很多不必要的分心。人的注意力很珍贵。

一个好的空间，应该让人把注意力用在交流、工作、用餐、思考、休息，而不是一直被光牵着走。

---

也正因为如此，好光最后带来的，从来不只是“看起来不错”，而是更深一层的结果：

人愿意停留。
愿意交流。
愿意消费。
愿意工作。
愿意放松。
愿意再回来。

这才是很多商业空间、服务空间，甚至居住空间真正该在意的价值。

---

四、照明真正该追求的，也许不是“惊艳”，而是“久待”。

这些年看了很多空间，也做了很多和光有关的事情，我越来越在意一件事：

一个空间是不是让人愿意久待，可能比它是不是第一眼惊艳，更重要。

因为“惊艳”是瞬间的。但“久待”，才真正关乎空间有没有服务到人。

---

对商业空间来说，久待意味着停留时间、体验品质、交流氛围，甚至最后的转化。
对酒店来说，久待意味着放松感、安定感、记忆点。
对办公空间来说，久待意味着疲劳控制、专注效率与长时间工作的舒适度。
对居住空间来说，久待则更直接，它关乎陪伴、节奏与生活本身。

---

所以我常觉得，照明真正成熟的一步，不是把光做得更“用力”，而是把光做得更“恰当”。

不是一味吸睛，而是更懂得停留。
不是一直强调存在感，而是懂得把空间还给人。
不是让每一处都在说话，而是让整体终于能安静下来。

---

这种转变，看起来不那么张扬，却可能是空间品质真正升级的开始。

---

你最近待过最舒服的一个空间，是哪里？

它未必最华丽，但很可能在光上做对了一些事。

The issue is not just the hardware. It is the DLMO logic.

Abstract

This article examines how LED component makers, luminaire manufacturers, control system providers, and lighting designers can build truly effective circadian lighting by following DLMO logic. It argues that the real challenge is not isolated parameters, but cumulative dose, eye-level delivery, and outcome validation.

Over the past few years, more and more companies have started talking about circadian lighting, sleep lighting, and healthy light.

But if we return to the underlying logic of DLMO — Dim Light Melatonin Onset — we quickly realize something important: truly effective circadian lighting is not simply about making light cooler during the day, warmer at night, or adding tunable white and dynamic scenes.

The real issue is this:

What kind of total light exposure does a person receive over the course of a day, at the eye, in the right timing windows, and does that exposure actually change physiology and behavior in a meaningful way?

That is why I increasingly believe that the next real competition in circadian lighting will not be about who can tune more parameters. It will be about who can build an integrated solution around hardware + scenes + cumulative dose + validation.

1. Why DLMO changes the definition of circadian lighting

DLMO matters because it does not simply describe whether someone sleeps well.

It helps answer a more fundamental question:

When does the body’s internal night actually begin?

In sleep medicine and circadian science, DLMO is widely used as a key phase marker of the central circadian clock, and it is often used to optimize the timing of bright light and melatonin interventions. Melatonin typically begins to rise about 2–3 hours before habitual sleep onset, and DLMO marks that transition.

This means circadian lighting should not be defined merely as “healthy-looking lighting.”

It should be judged by whether it can answer questions like:

• Does it provide enough effective circadian stimulus during the day?
• Does it reduce stimulation at the right time in the evening?
• Does it avoid suppressing melatonin when the body is preparing for sleep?
• Does it help stabilize or shift circadian phase in the intended direction?

In other words, the true objective is not a single snapshot parameter.

It is the daily exposure trajectory.

I prefer to summarize this in two words: cumulative dose.

2. The most common reason circadian lighting fails: it ignores cumulative dose

When companies discuss circadian lighting, the first things they usually mention are:

• spectrum
• CCT
• dynamic dimming
• pre-set scenes

All of these matter. But on their own, they are not enough. The circadian system does not decide its response from a single glance.

It is shaped by accumulated time cues across the day:

• Was morning light strong enough and early enough?
• Was daytime exposure sustained and effective?
• Did stimulation drop at the right time in the evening?
• Was night-time exposure sufficiently reduced, or was the system repeatedly disturbed?

The 2022 expert recommendations in PLOS Biology clearly state that healthy adults should receive relatively high melanopic EDI during the day, significantly lower levels in the 3 hours before bedtime, and as little as possible during sleep. These recommendations are based on vertical eye-level exposure, not just workplane illuminance.

The industry implication is clear: Circadian lighting cannot stop at “what is the fixture doing right now?”

It also has to answer:

• How much effective circadian stimulus did this user actually receive today?
• In which timing windows did it occur?
• Was there unnecessary stimulation at the wrong time?
• Is the total exposure profile supporting entrainment, phase advance, maintenance, or disruption?

If a company cannot answer these questions, then many so-called circadian lighting solutions are still just adjustable white lighting.

3. What LED component makers need to upgrade first

Some people think DLMO is far removed from LED package manufacturers.

I think the opposite.

If the upstream logic does not evolve, the downstream ecosystem will struggle to build truly effective circadian solutions.

1) Move from “efficacy + CCT + CRI” to a dual visual + non-visual language

Most LED data sheets still focus on:

• efficacy
• CCT
• 比显指
• binning
• lifetime
• electrical performance

These remain essential. But they are no longer sufficient for the circadian era.

Upstream suppliers should increasingly provide:

• spectral power distribution
• melanopic / α-opic relevant quantities
• non-visual performance under different spectral mixes
• spectral stability under dimming
• circadian consistency under different drive conditions

Because designers and control platforms are no longer just creating white light.

They are building time-based light recipes.

If LED suppliers cannot provide stable, traceable, and model-friendly spectral information, downstream players will struggle to implement circadian strategies with precision.

2) Move from “one universal LED” to “programmable spectral capability”

The future is not about one fixed optimum point. It is about spectral combinations that can be shifted over time, with predictable non-visual impact.

That means component makers should begin thinking in terms of:

• spectral mixes suited for morning phase-advancing stimulus
• daytime performance-supportive stimulus
• evening wind-down modes
• low-disruption night pathways

That is no longer the traditional logic of selling a light source. It is the beginning of selling a programmable circadian foundation.

4. What luminaire manufacturers must really do: deliver dose to the eye

Circadian lighting does not happen on a specification sheet. It happens at the human eye.

So for luminaire manufacturers, the real challenge is not just enabling tunable white.

It is ensuring that the intended dose reaches the eye in a controllable and useful way.

1) Move from plane-based lighting to eye-level lighting

Circadian effectiveness is more closely related to actual eye exposure than to horizontal workplane illuminance. The major recommendations increasingly emphasize melanopic exposure at the eye.

This means luminaire design must increasingly consider:

• optical direction
• emitting surface position
• the balance between glare control and effective circadian delivery
• direct / indirect / semi-indirect proportions
• seated, standing, and reclined eye positions

简而言之： The goal is not just to illuminate the room. It is to deliver the right circadian dose to the eye.

2) Move from one luminaire logic to time-based luminaire roles

A more mature circadian lighting system may not rely on one luminaire doing everything.

Instead, it may include:

• stronger morning/daytime stimulus luminaires
• transitional evening luminaires
• low-disruption night luminaires
• bedroom or hotel pathway lighting
• dual-logic luminaires for care tasks versus rest protection

This is especially important in real-world environments. Research in ICU settings shows that visual task needs and circadian goals are not naturally aligned, and that dynamic, zoned, and time-based solutions are more realistic than static ones. 

For luminaire companies, this means product families should move from “selling by room type” to “selling by time-task logic.”

5. Why control systems matter more than ever

I have said this for years: in the HCL and circadian era, control systems will be revalued.

Because circadian lighting is not fundamentally a static hardware problem.

It is a time orchestration problem.

1) Move from scene switching to circadian scripting

Traditional control systems are mostly valued for:

• scheduling
• dimming
• CCT control
• occupancy sensing
• energy savings
• scene recall

But once we apply DLMO logic, the system has to answer more:

• When should the morning stimulus begin?
• How quickly should it ramp?
• How should daytime cumulative dose be maintained?
• When should evening reduction begin?
• How should night-time visual needs be preserved while minimizing circadian disruption?

This is no longer just “having scenes.”

It is building physiology-aware time scripts.

2) Control systems must begin to account for cumulative exposure and feedback

This is one of the biggest future dividing lines.

Advanced circadian controls should not only know the current output value.

They should increasingly be able to:

• estimate cumulative effective exposure over time
• adjust electric light based on daylight contribution
• respond to occupancy and activity type
• estimate real user exposure
• calibrate strategy with measurements and feedback

In other words, the system should know more than “the room is currently 4000 K and 300 lux.”

It should increasingly understand: How much useful daytime signal has this person already received today, and how much circadian margin remains for the evening?

That is a much more DLMO-aligned system logic.

6. Designers are becoming “time experience orchestrators”

If LED component makers define the spectral raw material, luminaires determine delivery, and control systems determine temporal behavior, then designers ultimately determine how all of this becomes human experience.

This is why I believe the role of the designer is being fundamentally upgraded.

1) Design must move beyond “how bright and how beautiful”

It must also ask:

• At what time should this space support alertness?
• At what time should it support restoration?
• At what time must stimulation be reduced?
• Do different users in the same space have different circadian needs?
• How do visual comfort, operations, maintenance, and circadian goals coexist?

That changes the design starting point.

2) Designers must move from parameter thinking to exposure trajectory thinking

The strongest designers in the next phase will not only specify:

• 3000 K / 4000 K
• 300 lx / 500 lx
• UGR / CRI

They will increasingly specify:

• 7:00–9:00: rapid morning stimulus build-up
• 10:00–15:00: sustained daytime signal
• after 18:00: marked reduction in eye-level melanopic exposure
• after 22:00: only low-disruption pathway lighting
• distinct strategies for bed, desk, social, washroom, and transit activities

That is much closer to a DLMO-aware design language.

7. Hardware and scenes must be designed together

Many companies still follow this sequence: First build the product.

Then look for a “healthy lighting” use case. In circadian lighting, that order is often backwards.

A more appropriate logic is: Define the scene objective first, then define the hardware requirement

Scene 1: Office

The goal is not simply high illuminance.

It is sufficient daytime stimulus, limited evening carry-over, and good visual performance.

That leads to hardware needs such as:

• strong morning/daytime eye-level exposure
• 日光协同
• low glare without losing useful stimulus
• spectral and control capability for evening reduction

Scene 2: Bedroom / Hotel

The goal is not to create a “sleep lamp” gimmick. It is to reduce melatonin suppression opportunities while preserving necessary function.

That leads to hardware needs such as:

• very low-disruption night lighting
• low-stimulus bathroom and pathway lighting
• evening transition modes
• morning wake-up modes

Scene 3: Healthcare / Senior living / Wellness

The goal is to balance care tasks, resident rest, and staff circadian needs.

That means hardware and controls must support:

• zoning
• scheduling
• role-based logic
• task-based logic
• traceable validation

So future circadian lighting products cannot be defined independently of scenes.

Hardware must be designed for the scene, and the scene must be supported by the hardware.

8. Why I keep emphasizing validation

Because one of the biggest risks in this field is that many claims sound persuasive, but the outcomes may not be real.

From a DLMO perspective, circadian lighting should not be judged only by whether it is dynamic.

It should be judged by whether it changes the intended outcome.

At least three layers need validation

1) Output validation

Is the system actually delivering what was designed?

• actual SPD
• actual illuminance
• actual eye-level exposure
• actual time profile
• actual stability across dimming and tuning

2) Dose validation

Did the user actually receive the intended cumulative dose?

• enough morning stimulus?
• enough daytime accumulation?
• timely evening reduction?
• sufficiently low night-time exposure?

3) Outcome validation

Did that exposure trajectory actually change anything meaningful?

• alertness
• comfort
• sleep quality
• circadian stability
• task performance in target applications

This is why the ICU dynamic lighting study is valuable. It did not stop at comparing lighting configurations. It also looked at visual comfort and biological markers such as melatonin and cortisol. The authors rightly note the sample-size limitations, but the direction is important:

circadian lighting ultimately has to return to outcomes. 

9. Why validation toolchains matter

If the industry is serious about cumulative dose and real outcomes, we can no longer rely only on intuition and one-off impressions.

That is one reason we have continued building the In.Licht toolchain.

In.Licht Pro

Better suited for field diagnosis, inspections, and practical project communication.

It helps teams see the basic light facts more clearly.

In.Licht Ultra

Better suited for R&D, quality control, comparison, and deeper spectral analysis.

It helps teams understand why two lights that look similar may behave very differently.

In.Licht Well

Better suited for continuous monitoring and operational management.

It helps move from “measure once” to long-term optimization, integrating EML, M-EDI, and broader environmental factors into one workflow.

Together, they support a much more useful workflow: see the facts → understand the mechanism → build cumulative dose logic → validate → optimize

⸻

10. Who will win next?

I increasingly believe that the winners in the next phase will not simply be the companies that talk most about HCL or healthy light.

They will be the ones that build real capability around:

• programmable and model-ready spectral foundations
• luminaires that deliver dose effectively to the eye
• control systems that orchestrate physiological timing
• design methodologies that integrate time, space, and activity
• field workflows that measure, validate, and improve continuously

At the center of this are two ideas:

cumulative dose

outcome validation

The companies that can build products and projects around those two ideas will be much closer to the real value of next-generation circadian lighting.

Conclusion: circadian lighting is not just another mode

If we rethink the industry through the lens of DLMO, we see that circadian lighting is not simply a new “sleep mode” added to conventional lighting.

It is a more fundamental shift:

• LED components become the foundation of temporal light recipes
• luminaires become dose delivery devices
• controls become time orchestration systems
• design becomes time experience design
• validation becomes outcome validation, not just brightness checking

That is how I understand circadian lighting.

And that is why I believe the real future value lies not in one isolated product, but in a methodology that integrates:

hardware + scenes + cumulative dose + effect validation

That, in my view, is one of the most important directions for the next upgrade of our industry.

Call-to-Action

如果你是

• an LED component maker
• a luminaire manufacturer
• a control system platform
• a lighting designer or consultant
• or a brand exploring circadian lighting for offices, hotels, residential, wellness, healthcare, or senior living

I would be glad to connect.

At 光配方研究院（Lighting Recipe Studio, LRS）, we are interested in working with partners on:

• DLMO-based product definition for circadian lighting
• scene scripting based on cumulative dose logic
• integrated R&D, measurement, and validation workflows
• joint development from concept to real-world deployment

Because circadian lighting should not just be dynamic.

It should be effective.